Fredmt Empowering Electric Vehicles and Sustainable Energy

Blog

Next-Generation Geothermal Technologies Are Heating Up

Beneath our feet lies a nearly unlimited source of energy, but while a few fortunate locations have geothermal heat near the surface, the majority of the world will need to dig much deeper. The challenge lies in how to reach adequate depths.

There are certain places globally where energy literally rises to the surface. In Iceland, which has over 200 volcanoes and numerous natural hot springs, accessing this energy is relatively easy. The country is dotted with steaming water pools, heated by geothermal activity just below the Earth’s crust. Geysers erupt boiling jets of water and steam into the air.

Iceland now utilizes geothermal energy to heat 85% of its homes, and 25% of the nation’s electricity is also derived from power plants that harness this underground heat. This presents an attractive opportunity—an almost limitless energy source waiting to be tapped into.

Geothermal energy provides an essentially inexhaustible green energy option worldwide. Furthermore, it is “always on,” unlike wind or solar energy, because the heat is perpetually emitted from the Earth’s molten core and the decay of naturally occurring radioactive materials in our planet’s crust. In fact, the Earth releases such massive amounts of energy as it cools that the heat lost into space annually could satisfy the world’s total energy requirements many times over. The challenge remains in how to access that energy.

At present, only 32 countries across the globe operate geothermal power plants. There are under 700 power plants worldwide, producing around 97 Terawatt hours (TWh) in 2023 collectively. This amount is less than half of the electricity generated by solar energy in the US alone and falls significantly short of projections for geothermal’s potential to contribute to the global energy landscape. Some estimates suggest that geothermal could generate approximately 800-1400 TWh of electricity annually by mid-century, along with an additional 3,300-3,800 TWh per year of heat.

“The Earth itself has the potential to tackle a variety of challenges in the transition to a clean energy future,” stated Amanda Kolker, geothermal programme manager at the National Renewable Energy Laboratory (NREL) in the US, when presenting a report on geothermal energy’s potential in 2023.

However, not all countries are as fortunate as Iceland, where reservoirs of hot water at temperatures between 120-240°C (248-464°F) can be easily accessed at shallow depths. In various regions of the country, wells drilled to depths of up to 1.5 miles (2.5 km) can reach temperatures as high as 350°C (662°F). For instance, Iceland’s main geothermal site at Reykjanes has drilled exploratory wells down to 2.9 miles (4.6 km) to access superheated fluids reaching up to 600°C (1112°F). Currently, geothermal energy extraction occurs using shallower wells that tap into temperatures around 320°C (608°F) to generate 720 Gigawatt hours (GWh) of electricity yearly.

One reason why geothermal energy is not more widely used is the significant initial investment needed for energy extraction. Additionally, physically reaching these depths has also presented challenges thus far.

For other regions of the world to benefit from this geothermal clean energy bounty, deeper drilling is essential to access the necessary temperatures for electricity generation or large-scale heating for nearby communities.

Across much of the globe, temperatures typically increase by 25-30°C (77-86°F) for every kilometer one descends through the Earth’s crust. For instance, in the UK, the subsurface temperature at roughly 5 km (3 miles) deep is about 140°C (284°F), according to the British Geological Survey.

However, if one digs deeply enough, it is feasible to reach a point where water temperatures exceed 374°C (705°F) at pressures greater than 220 bars (with one bar representing average pressure at sea level). This temperature and pressure combination leads to an energy-intensive state known as supercritical, where water exists in a form that is neither liquid nor gas. The hotter and more pressurized it becomes, the more energy it holds.

In fact, a single superhot geothermal well could produce five to ten times the energy generated by current commercial geothermal wells, according to the NREL.

One significant obstacle, however, is that conventional rotary drills—even those equipped with diamond tips—are poorly suited for reaching the depths required to access these temperature levels. In the enigmatic deep underworld, characterized by uncertain geology, extreme temperatures, and immense pressures, drill components frequently fail, and preventing holes from becoming blocked presents a constant struggle.

In 2009, a team involved in the Iceland Deep Drilling Project accidentally reached supercritical conditions when they drilled into a magma chamber located roughly 1.2 miles (2km) beneath the surface at the Krafla volcano. The steam released from this well was extremely hot and acidic, complicating its usability. The intense pressures and temperatures made it challenging to manage, requiring intermittent discharges for about two years until a valve failure led to sealing the hole.

Deep drilling is often a costly and time-intensive process

The deepest hole ever bored by humans dates back to the Cold War, during which there was competition between superpowers to drill as deeply as possible into the Earth’s crust. The Soviets succeeded in drilling through 7.6 miles (12.2 km) of rock, creating the Kola Superdeep Borehole on the Kola Peninsula in the Arctic Circle. They spent nearly two decades reaching that depth, which still stands as the greatest depth humans have reached in the Earth. (For more on the Kola Superdeep Borehole, see this article by Mark Piesing.)

The National Renewable Energy Laboratory (NREL) estimates that drilling a 1 km deep well costs about $2 million (£1.57 million), while drilling four times that depth can range from $6 million to $10 million (£4.7 million to £7.87 million) with current technology.

However, deep geothermal energy has the potential to offer significant cost savings in comparison to traditional geothermal systems, thanks to the elevated temperatures and pressures accessible deeper in the Earth’s crust. Some research indicates that deep geothermal energy could provide heating for communities at prices comparable to other heating methods, such as gas, but with reduced greenhouse gas emissions.

With this in mind, innovative researchers and companies are exploring new drilling methods and technologies to create some of the deepest holes ever in order to harness geothermal energy in regions previously thought unsuitable.

Quaise Energy, a spin-off from the Massachusetts Institute of Technology (MIT), aims to drill to depths of 12 miles (20 km) to access temperatures exceeding 500°C (932°F). They are employing a tool based on years of study into nuclear fusion technology. “While others are using traditional drills, we are introducing microwaves into the ground for the first time,” states co-founder Matt Houde.

He and his team are testing millimeter-wave directed energy beams that can vaporize even the toughest rock. This technology directs a high-powered radiation beam, similar to microwaves but at a higher frequency, onto a rock segment, heating it up to 3,000°C (5,432°F) to melt and vaporize it. By directing the beam to penetrate the rock, it allows for hole creation without the debris and friction associated with conventional drilling methods.

“Millimeter-wave drilling can function largely independently of depth,” Houde explains. “Additionally, this millimeter-wave energy can pass through dirty and dusty environments.”

This technology is derived from nuclear fusion plasma experiments conducted by Paul Woskov, an engineer at MIT’s Plasma Science and Fusion Center. Since the 1970s, millimeter-wave directed energy has been investigated as a means to heat plasma in nuclear fusion reactors, but a few years ago, Woskov discovered an alternative application for the technology and began using millimeter-wave beams produced by a gyrotron to melt through rock.

So far, the technology has only undergone laboratory testing, achieving shallow drilling in relatively small rock samples; however, the company claims a drilling rate of about 3.5 m (11.5 ft) per hour. Although this is slower compared to traditional methods, it offers other advantages, as the “drill bit” does not physically grind through the rock, so it should not wear down or require replacement. Quaise Energy is currently in the final stage of lab testing for millimeter-wave technology and plans to commence field trials in early 2025.

Nonetheless, transferring millimeter-wave drilling technology from the lab to full-scale operations presents challenges.

“They have never been utilized in the deep high-pressure subsurface environment before,” Woskow comments. “The changes resulting from intense energy-matter interaction during drilling necessitate a new learning curve.”

A Slovakia-based company, GA Drilling, is investigating an alternative high-energy drilling technology designed to penetrate the Earth’s crust. They are employing a pulse plasma drill that utilizes very brief, high-energy electric discharges to break apart rock without melting it. This method prevents the creation of thick molten rock, which can be challenging to remove and may hinder further drilling. “As the process rapidly disintegrates the rock with brief shocks, there isn’t enough time for melting to occur—thus, the frequency of needing to pull up and change the drill bit is significantly lower,” states Igor Kocis, the chief executive and chairman of GA Drilling. “Our current development program aims for depths of five to eight kilometers (3-5 miles)—and eventually over 10 kilometers,” he adds. “These depths will provide nearly universal access to geothermal energy.”

Research into pulse plasma drills, which use extremely short energy pulses to fragment rock with ionized gas reaching temperatures of up to 6,000°C (10,832°F), is being pursued by a European consortium led by the Geothermal Energy and Geofluids (GEG) group, along with partners from Germany and Switzerland.

GA Drilling has also collaborated with Konstantina Vogiatzaki, an associate professor of engineering science at the University of Oxford, to apply advanced mathematical techniques to manage supercritical fluids when accessing deep Earth energy sources through plasma drilling. “We focused on determining the ideal combustion system for a full-scale drilling tool, paving the way for better control of ultra-high pressure combustion via plasma drilling,” Vogiatzaki explains.

Others are exploring methods beyond our own planet to facilitate deeper drilling. Technologies initially developed for planetary exploration missions on the intense surface of Venus, where temperatures can soar to 475°C (887°F), are being repurposed by companies in the geothermal drilling sector. Ozark Integrated Circuits, an electronics firm located in Fayetteville, Arkansas, has been modifying circuits to endure extreme temperatures, suitable for use in deep Earth geothermal drilling rigs.

In its efforts, the U.S. National Renewable Energy Laboratory (NREL) has implemented artificial intelligence to analyze complex subterranean conditions to identify optimal drilling sites for supercritical water and to help foresee and recognize issues with drills before they escalate into significant problems.

Some companies are already advancing in deep Earth exploration. The geothermal firm Eavor stated to the BBC that in 2024, it achieved a depth of three miles (5 kilometers) with two vertical wells at a site in Geretsried, Bavaria, Germany. It is utilizing two of the largest land-based drilling rigs in Europe to develop a commercial-scale facility in Geretsried that aims to extract geothermal heat by circulating water in a closed-loop system referred to as the Eavor Loop. This system operates similarly to a large radiator, where cold water in the loop is heated underground and then returned to the surface for electricity generation and distribution to nearby homes through a district heating network. Eavor anticipates beginning energy production at the site in early 2025, according to John Redfern, the company’s CEO and president.

“Our technology aims to reach drilling depths of up to 11 kilometers (6.8 miles) in the future,” remarks geologist and co-founder of Eavor, Jeanine Vany. “I am confident we can make significant strides in tapping superhot rock within the next three to five years.”

The closed-loop design also helps mitigate some contamination issues associated with extracting superheated water from deep geothermal wells, as highlighted by the Iceland Deep Drilling Project in 2009. It may also reduce emissions of harmful gases such as hydrogen sulfide, which can be a byproduct of open-loop geothermal systems.

Vany further emphasizes that deep geothermal energy requires minimal surface space, indicating that it could be integrated into urban settings in the future.

However, there are additional challenges that need to be addressed. It remains uncertain how simple it will be to maintain deep geothermal wells and prevent them from becoming obstructed.

The pursuit of deep geothermal energy may also revitalize aging fossil fuel power plants, as nations aim to phase out their traditional carbon-emitting energy sources. Retrofitting old coal power plants into geothermal facilities could provide these steam-powered generators with a renewed purpose and facilitate the swift establishment of geothermal plants by leveraging existing electricity transmission infrastructure. Woskov has pinpointed an unused coal power facility in upstate New York, which he hopes to reactivate before the decade concludes, to generate electricity from subsurface heat.

There is a certain poetic element to this transition—a power station that once operated on a polluting fuel extracted from the earth finding new vitality in the clean energy movement with a source from deeper beneath the surface. The looming question remains—will they manage to drill deep enough?

December 3, 2024
The New York Attorney General’s office fined car insurance company Geico $9.75 million

The New York Attorney General’s office imposed a fine of $9.75 million on Geico for hacks that compromised the personal information of 116,000 drivers in the state. The Attorney General and the state Department of Financial Services stated that both Geico and Travelers Indemnity Company breached state data protection regulations by inadequately implementing measures to safeguard consumers’ information.

Both firms were targeted by hackers during the COVID-19 pandemic, amidst a surge of cyberattacks aimed at extracting details such as drivers’ license numbers for fraudulent unemployment claims, according to the agencies. Travelers will incur a penalty of $1.55 million for a security breach that revealed information on about 4,000 individuals, as reported by the agencies.

Both companies have agreed to take steps to enhance their cybersecurity protocols. A representative from Geico mentioned that the company reported the incident to the state and has since allocated significant resources toward bolstering its cybersecurity defenses. A spokesperson for Travelers has not yet responded to a request for comment.

In a noteworthy enforcement action, New York Attorney General Letitia James has levied a joint fine of $11.3 million against the insurance giants Geico and Travelers Indemnity Company for data breaches that placed the personal information of over 120,000 people at risk during the COVID-19 pandemic. The penalties, disclosed by the New York Department of Financial Services (DFS), underscore serious deficiencies in the cybersecurity practices of both companies, which were exploited to steal sensitive information such as drivers’ license numbers and personal data.

The data breaches affecting Geico and Travelers highlighted security vulnerabilities that, although typical in cyber incidents, emphasize areas where enhanced measures could have potentially reduced the risk.

Geico’s breach originated from weaknesses in its online quoting tool—a system designed to make acquiring insurance quotes easier for customers. Between 2020 and 2021, attackers took advantage of this tool through credential stuffing attacks. In this method, cybercriminals utilized stolen usernames and passwords from earlier data breaches, testing various combinations until they eventually gained access.

After breaching the system, attackers managed to extract the drivers’ license numbers of around 116,000 individuals. While this information isn’t a direct financial target, it can play a critical role in identity theft schemes, such as filing fraudulent unemployment claims, a problem that increased during the pandemic.

This breach highlights the necessity of implementing protections like CAPTCHA and other automated bot detection mechanisms in systems that handle sensitive information. Strengthened verification measures, such as multi-layered identity checks, could have provided additional protection against these types of assaults.

The Travelers data breach occurred in April 2021 and compromised the data of about 4,000 individuals. The attackers gained entry by utilizing stolen employee credentials, a technique that circumvented the company’s defenses because of the lack of multifactor authentication.

MFA, which necessitates users to confirm their identities using a secondary factor, such as a code generated on a mobile phone, is regarded as a fundamental security measure in today’s threat environment. In the absence of this barrier, the attackers were able to access the system using just a username and password.

Although there have been no reports of misuse regarding the exposed data, this incident illustrates the critical need for adopting MFA as standard practice to protect internal systems. Both breaches took place during a time of increased online activity fueled by the COVID-19 pandemic, demonstrating how attackers exploited weakened systems and widespread remote work to capitalize on known vulnerabilities.

The fines—$9.75 million for Geico and $1.55 million for Travelers—underscore New York’s role as a frontrunner in cybersecurity regulation. The DFS enforces stringent requirements under its Cybersecurity Regulation, 23 NYCRR Part 500, which mandates financial organizations to uphold robust cybersecurity programs, regularly evaluate risks, and implement safeguards such as MFA.

Both companies were found in violation of these regulations: Geico’s inability to secure its online quoting tool permitted unauthorized access to sensitive customer information, while Travelers’ omission of MFA rendered internal systems susceptible to breaches.

“DFS’s pioneering cybersecurity regulation provides an essential framework for ensuring the protection of sensitive consumer data and the resilience of financial institutions,” remarked New York State Financial Services Superintendent Adrienne Harris. “These enforcement actions reinforce the Department’s commitment to ensuring that all licensees, especially those responsible for consumer financial information like GEICO and Travelers, fulfill their obligation to implement strong measures that protect New Yorkers from potential data breaches and cyber threats. I appreciate the collaboration with the Attorney General’s office during these efforts.”

In response to my inquiry, a representative from Geico stated, “GEICO is happy to have come to a resolution regarding this issue with the New York State Department of Financial Services and the New York State Attorney General. Upon discovering this problem, GEICO voluntarily reported it to officials in New York State and made enhancements to its systems to avert further exploitation by these fraudsters. GEICO is serious about data security and has made substantial commitments to bolster its cybersecurity efforts.”

Consequences for Consumers: The Lasting Effects of Data Breaches

For those affected by the breaches at Geico and Travelers, the repercussions extend beyond the initial exposure of personal information. The aftermath can influence financial security and long-term stability.

The Financial Ramifications of Data Breaches

In the case of Geico, the theft of driver’s license numbers created avenues for criminals to submit fraudulent unemployment claims. These scams not only interfered with legitimate claims but also compelled affected individuals to dedicate considerable time and effort to verifying their identities and contesting false submissions. In some instances, these fraudulent claims may have postponed crucial benefits for victims during an already difficult period.

For Travelers, while fewer individuals were involved, the breach revealed personal information that could facilitate identity theft or other fraudulent activities. The revelation of such data adds a layer of apprehension for those impacted, even if there’s no immediate report of misuse.

The Emotional and Pragmatic Impact of Data Breaches

Beyond the financial consequences, the emotional strain on victims is considerable. The awareness that personal information is in the possession of unknown individuals results in anxiety and a persistent feeling of vulnerability. Victims often find themselves questioning how, when, or if their data could be exploited in the future.

Recovering from these breaches can take a significant amount of time. Victims might need to keep an eye on their credit for suspicious activities, place fraud alerts or freezes on their accounts, and consider investing in identity protection services. This process often entails not just addressing immediate concerns but also remaining alert for potential future abuse of stolen information.

Data Breach Enforcement Intensifies

The breaches at Geico and Travelers underscore the extensive ramifications of data exposure, impacting not only companies but also the individuals whose personal details have been compromised. New York’s regulatory actions indicate that authorities are increasingly holding organizations accountable for safeguarding sensitive information, highlighting a larger movement towards enhanced cybersecurity practices across various sectors. For consumers, these incidents serve as a reminder to remain vigilant in monitoring their accounts and protecting their personal information.

Both Geico and Travelers have been approached for comments. Geico provided a statement that has been integrated into the article.

The DFS outlined that Geico and Travelers had insufficient security measures, resulting in the compromise of sensitive information. The breaches involved a sequence of cyberattacks targeting Geico starting in 2020 and one against Travelers in 2021.

In both scenarios, attackers accessed the companies’ third-party auto insurance quoting tools and extracted driver’s license numbers.

The DFS asserted that Geico did not adequately secure its publicly accessible website and neglected to thoroughly review its systems after being informed about the attack campaign. Although Geico addressed vulnerabilities affecting its website, attackers managed to exploit weaknesses in Geico’s insurance agents’ quoting tool to access the data. This attack compromised the driver’s license numbers of 116,000 residents of New York State.

Concerning Travelers, the DFS stated that the insurer failed to adopt adequate security protocols despite prior warnings of an attack campaign targeting insurance quoting tools. During the assault on Travelers, threat actors used stolen credentials belonging to Travelers agents. The DFS noted that Travelers’ agent portal lacked multi-factor authentication, which attackers exploited to gain initial entry.

The DFS reported that it took more than seven months for Travelers to identify suspicious activity on the compromised agent portal. This attack exposed the personal information of 4,000 individuals from New York.

Due to the breach, Geico was fined $9.75 million, and Travelers faced a penalty of $1.55 million. Moreover, both insurers are obligated to enhance their security protocols by creating and maintaining a data inventory of private information, strengthening threat detection and response tools, and improving authentication methods.

In the consent order involving the DFS and Geico, the agency revealed that attackers took advantage of a vulnerability found in the third-party quoting tool 75 times between 2020 and 2021. It also disclosed that attackers demanded ransom from Geico during the 2020 incident.

“Geico did not identify the Third Cybersecurity Event until March 1, 2021, when it received messages from threat actors trying to ransom stolen customer data back to Geico, along with separate communications from an individual detailing a personal dispute with the threat actors and guiding GEICO on the precise steps taken to steal the customer data and what actions GEICO needed to implement to address the vulnerability,” New York State DFS mentioned in the consent order.

The consent order highlighted further aspects where Geico’s security was deficient. For instance, it stated that Geico failed to encrypt sensitive information or carry out annual penetration tests on its network.

Geico was mandated to perform a cybersecurity risk assessment within 30 days following the issuance of the consent order.

TechTarget Editorial reached out to Travelers for a response, and a company representative provided this statement:

“We are glad to have settled this issue, which involved the compromised credentials of a limited number of independent agents. Safeguarding the information of all our stakeholders is a top priority, and we will continue collaborating with our independent agents to thwart similar incidents in the future. It is vital to emphasize that Travelers’ internal systems were not affected by this incident.”

Geico and Travelers are among the latest firms to face fines from regulatory authorities due to security deficiencies this year. Last month, the U.S. Federal Trade Commission instructed Marriott International Inc. to pay $52 million in fines and enhance their security measures after three data breaches impacted over 300 million customers. T-Mobile was also ordered to pay a $15.75 million penalty last month by the Federal Communications Commission concerning the telecom giant’s management of various data breaches. As part of the settlement, T-Mobile is required to invest another $15.75 million into its enterprise security program.

GEICO and Travelers are two of the largest auto insurance providers in the United States, and if you’re looking for new coverage, you’re likely comparing these companies.

GEICO ranks as the third-largest auto insurance company in America and provides coverage for over 28 million vehicles. The company has received high ratings for financial stability and an Insurify Quality (IQ) Score of 9.0 due to its competitive rates, round-the-clock customer service, and various discounts available.

Travelers boasts strong financial strength ratings and has a 165-year history of providing automobile insurance. Travelers also earned an IQ Score of 9.0 and has received high ratings for customer service in certain markets. However, online customer reviews are mixed.

GEICO vs. Travelers: The conclusion

GEICO car insurance tends to offer lower premiums compared to Travelers for both full coverage and liability-only policies. Drivers without a previous accident, those who have had an accident, or individuals with a DUI will generally find that GEICO’s average car insurance rates are more economical than those of Travelers.

GEICO auto insurance also surpassed Travelers in most regions for customer satisfaction in the 2022 J.D. Power U.S. Auto Insurance study, and it holds an A+ rating with the Better Business Bureau (BBB), whereas Travelers has an A rating.

However, for drivers seeking to work with a licensed insurance agent, GEICO has a considerably smaller network, with approximately 300 agents and brokers nationwide, in contrast to 13,500 for Travelers.

GEICO

GEICO ranks as the third-largest auto insurance provider in the U.S. and has received favorable evaluations from both the BBB and in J.D. Power’s U.S. Auto Insurance Study.

GEICO, a subsidiary of the Berkshire Hathaway Group, is the third-largest insurer in the U.S. based on market share. In addition to auto insurance, GEICO offers homeowners, renters, flood, travel, life, and business insurance, among other products. The insurer presents various car insurance discounts to assist drivers in saving money. For instance, drivers who maintain an accident-free record for five years can receive a 22% discount on premiums, while good students may qualify for a 15% discount. GEICO’s complaint score from the National Association of Insurance Commissioners is below average, which indicates it receives fewer complaints from consumers than average.

Travelers

Travelers is a national insurance provider with an expansive network of agents that performed better than GEICO in the mid-Atlantic region as per J.D. Power’s U.S. Auto Insurance Study.

Travelers is the sixth-largest insurer in the U.S. in terms of market share and has provided services for more than 150 years. The company offers auto, homeowners, renters, flood, pet, and other insurance types. Besides standard auto coverage options such as liability, collision, and comprehensive coverage, Travelers provides gap insurance, accident forgiveness, new car replacement, and a variety of additional coverages. Drivers can benefit from a range of discounts, including those for bundling policies or covering a new vehicle. Travelers’ IntelliDrive program rewards policyholders with discounts of up to 30% for safe driving.

December 2, 2024
The Xiaomi SU7 Ultra received 3,680 pre-orders 10 minutes after pre-sales began
Xiaomi’s first electric vehicle, the SU7, has become an overwhelming success for the Chinese smartphone company, but it continues to demonstrate its ability to create a buzz in the domestic market.

During an annual conference in July, Xiaomi’s founder, chairman, and CEO, Lei Jun, introduced the SU7 Ultra prototype and has frequently highlighted its performance ambitions on China’s top social media platform, Weibo.

At the challenging Nurburgring in Germany, recognized as the global standard for performance car lap times, the four-door electric vehicle achieved a remarkable time of 6:46.874 – an astonishing 20 seconds quicker than the Porsche Taycan Turbo GT.

This represents a significant challenge for any manufacturer, especially for a company more accustomed to smartphones and consumer electronics.

As stated by Lei Jun, the record was accomplished with British endurance driver David Pittard at the wheel in just one lap – a remarkable single-attempt feat, so to speak.

In the footage from the onboard camera, which can be accessed on YouTube (or simply click above), the SU7 Ultra experienced a temporary power loss about two-thirds into the run. If this issue hadn’t occurred, the lap time could have been even more impressive.

Performance is made accessible

Not merely a publicity stunt, the Xiaomi SU7 Ultra, boasting 1,526hp and a tri-motor setup, is set to enter production early next year, joining the rest of the vehicle line-up. The price is anticipated to be 814,900 Chinese yuan – roughly $114,000/£88,000/AU$174,000 – but like the entire Xiaomi EV collection, it will only be available in China.

This vehicle offers remarkable value for the price, with specifications that read like a car enthusiast’s wish list: carbon ceramic brakes, a 0-62mph acceleration time of 1.98 seconds, a top speed of 217mph, adjustable Bilstein suspension, and an interior lavishly adorned with carbon fibre and Alcantara.

Although the production version will not be as extreme as the car that tackled the Nurburgring, it will still sport a rear spoiler, an updated front splitter with large air intakes, and an active rear diffuser for enhanced downforce.

It is perhaps unsurprising that Xiaomi received 3,680 deposits within the first 10 minutes of the car’s announcement, according to CarScoops.

Xiaomi’s SU7 Ultra secured 3,680 pre-orders just 10 minutes after pre-sales commenced as the company begins demonstrating its technological strength in the EV sector.

Xiaomi (HKG: 1810, OTCMKTS: XIACY) has introduced the production model of its SU7 Ultra, further showcasing its technical capabilities in the electric vehicle arena.

At a product launch event that featured announcements around smartphones, smartwatches, TVs, and washing machines, Lei Jun, the founder, chairman, and CEO, revealed that the powerful SU7 Ultra is now available for pre-order in China at a price of RMB 814,900 ($114,200).

The official launch event for the SU7 Ultra – which is priced nearly four times higher than the standard SU7, starting at RMB 215,900 – will take place in March 2025.

Customers can begin pre-ordering the SU7 Ultra immediately with a deposit of RMB 10,000, which is refundable at any time before the official release.

The vehicle received 3,680 pre-orders within just 10 minutes after the pre-sales began, as announced by Xiaomi’s automotive division, Xiaomi EV, on Weibo.

Characterized by Lei as Xiaomi’s “Dream Car” for himself and performance-oriented enthusiasts, the SU7 Ultra stands out as the most potent variation of the company’s first electric vehicle.

The SU7 electric sedan launched on March 28, featuring three variants – Standard, Pro, and Max – with starting prices of RMB 215,900, RMB 245,900, and RMB 299,900, respectively.

This month alone, over 20,000 units have been delivered, marking the first time this milestone has been achieved, as stated by Xiaomi during today’s event.

During his annual presentation on July 19, Lei introduced the SU7 Ultra prototype, noting that it was designed based on the SU7 for both performance and track use.

Xiaomi sought to challenge the renowned Nurburgring track with the SU7 Ultra prototype earlier this month, but the initial plan was postponed due to rain.

The prototype successfully completed that challenge yesterday, achieving a final lap time of 6 minutes and 46.874 seconds, making it the fastest four-door vehicle to lap the track, Xiaomi announced earlier today.

Xiaomi refined the chassis system for the production SU7 Ultra at the Nürburgring and began validation testing on June 11, according to Lei.

The model is designed for track use and can be operated on the course without modifications, as per the company’s claims.

Xiaomi’s tests indicated that the cooling system of the SU7 Ultra production car did not overheat after two laps at the Nürburgring.

The production version of the SU7 Ultra is equipped with three motors, including two Xiaomi V8s motors, each capable of generating 578 horsepower, along with a V6s motor that delivers 392 horsepower.

This powertrain offers a total output of up to 1,548 hp, enabling the SU7 Ultra to accelerate from 0 to 100 km/h in 1.98 seconds and achieve a top speed exceeding 350 km/h, thereby establishing it as the fastest four-door production vehicle.

The vehicle can go from 0 to 200 km/h in 5.85 seconds and reach 400 km/h in 9.23 seconds.

Constructed on an 800 V high-voltage foundation, this car features a 5.2 C charging multiplier, allowing it to charge from 10 percent to 80 percent in just 11 minutes.

The price for the production version of the SU7 Ultra aligns closely with the projections made by certain Wall Street analysts.

In a report earlier this month, Goldman Sachs estimated that Xiaomi’s SU7 Ultra will sell around 4,000 units next year, with an expected average price of about RMB 800,000, accounting for 5 percent of its electric vehicle revenue.

Xiaomi SU7 Ultra comes with the same tri-motor layout and a battery pack optimized for track performance as the Xiaomi SU7 Ultra Prototype, offering a peak power of 1548PS. It can accelerate from 0 to 100km/h in just 1.98 seconds (without one foot rollout) and has a maximum speed designed to reach 350km/h, making it the fastest four-door mass-produced vehicle. The model is equipped with a cooling system optimized for track use, ensuring that it can complete two laps on the Nürburgring Nordschleife without overheating.

Additionally, it features a top-tier braking system, achieving a stopping distance of only 30.8 meters from 100 km/h to a complete stop. Xiaomi SU7 Ultra is also provided with a peak chassis system fine-tuned for the Nürburgring Nordschleife, which enhances chassis control and raises the control limits. As a four-door “race car,” it is ready for track use right off the production line. Furthermore, the Xiaomi SU7 Ultra boasts various upgrades in areas like smart driving, smart cockpit, safety, and a luxurious experience.

On October 28 (German time), the Xiaomi SU7 Ultra Prototype accomplished its first lap challenge at the Nürburgring Nordschleife, breaking the long-standing seven-year record for the fastest four-door sedan with a time of 6’46″874. This marks the first instance of a Chinese brand earning the title of “The Nürburgring Nordschleife World’s Fastest Four-Door Car.” Xiaomi EV will continue to pursue innovative breakthroughs in electric technology and validate its vehicles through testing at the Nürburgring Nordschleife. The mass-produced array of Xiaomi SU7 Ultra is also set to take on the Nürburgring Nordschleife next year.

The Xiaomi SU7 Ultra is now open for pre-orders, with its official launch planned for March 2025. The pre-order price is set at 814,900 yuan, and a deposit of 10,000 yuan is required to express intent. Those who pre-order will receive priority delivery.

Designed for high performance: From streets to tracks, where aesthetic meets functionality

The design of the Xiaomi SU7 Ultra emphasizes optimal performance, with each new element serving a functional role, devoid of any unnecessary ornamentation. For its exterior, Xiaomi SU7 Ultra evolves from the Xiaomi SU7 Max, featuring an enhanced aerodynamic kit and larger body dimensions. With a length of 5115mm, width of 1970mm, height of 1465mm, and a wheelbase of 3000mm, the vehicle is longer and lower, exhibiting a more aggressive, battle-ready profile.

At the front, the Xiaomi SU7 Ultra showcases an oversized splitter and air dam, along with “U-shaped” air curtains that effectively boost downforce at the front end. A larger opening for the air intake grille has been introduced, increasing the heat dissipation surface area by 10%. An adaptive active rear diffuser has been incorporated, featuring two-speed adjustments that balance wind resistance, downforce, and energy efficiency for day-to-day driving. Additionally, a new carbon fiber fixed rear spoiler (with a wingspan of 1560mm and chord length of 240mm) generates substantial downforce at high velocities. Thanks to this new aerodynamic framework, the Xiaomi SU7 Ultra achieves a maximum downforce of 285kg, comparable to that of supercars.

The interior of the Xiaomi SU7 Ultra highlights sportiness, with improved seat and steering wheel designs. The newly designed sports seats offer enhanced side support, with quicker response times and broader support areas, enabling better body control during high-intensity driving. The seats also display exclusive track-inspired embroidery and an Ultra logo, adding to the sporty look.

The steering wheel has undergone a complete redesign, utilizing a race-inspired flat top and bottom. The material has been upgraded to carbon fiber, with the grip covered in Alcantara®️ microfiber to prevent slipping easily. A yellow centering marker at the 12 o’clock position assists with steering during vigorous handling. This new aesthetic, combined with yellow seat belts and a red Boost button on the steering wheel, creates a race-ready cockpit ambiance within the Xiaomi SU7 Ultra.

The interior materials of the Xiaomi SU7 Ultra have been further improved, extensively featuring Italian Alcantara®️ microfiber across the vehicle, covering over 5m². This material wraps around all seat contact areas, steering wheel grip zones, dashboard stitching, and interior door panels, providing a premium sensation and a more immersive driving atmosphere.

The Xiaomi SU7 Ultra extensively incorporates carbon fiber materials, enhancing its sporty feel while reducing overall weight. Inside, carbon fiber elements can be found in the front-seat back panels, center console, and door sills, which accentuate its athletic look. On the exterior, the Xiaomi SU7 Ultra showcases a large carbon fiber roof of 1.7m²*, cutting down weight by 12kg. Carbon fiber is utilized in 17 different locations, amounting to a total of 3.74m². Additionally, 90% of these components are processed using a high-quality hot press method, ensuring excellent texture and quality.

In terms of peak performance, it is recognized as the world’s fastest mass-produced four-door sedan around the Nürburgring Nordschleife, boasting 1548PS and achieving a 0-100km/h time of 1.98 seconds, with a top speed designed to reach 350km/h.

As Xiaomi’s premier high-performance electric vehicle, the SU7 Ultra features the same power system as its prototype, enabling outstanding performance.

The Xiaomi SU7 Ultra produces a peak output of 1548PS and can go from 0 to 100km/h in only 1.98 seconds (without one-foot rollout). It sets a new standard as the first mass-produced four-door sedan to accelerate from 0 to 200km/h in just 5.86 seconds and can achieve a designed top speed of 350km/h.

The exceptional performance of the Xiaomi SU7 Ultra is backed by leading-edge technology. It utilizes a tri-motor layout that includes two V8s e-motors and one V6s e-motor, marking the first commercial production of Xiaomi’s in-house developed HyperEngine V8s. This HyperEngine V8s can reach a maximum revolution of 27,200rpm, making it the most powerful and highest-revolution main drive e-motor currently on the market.

Xiaomi EV has continuously prioritized research and development (R&D) funding and maintained a focus on innovation. To date, the company has filed for 242 patents related to electric motors and electric power control units, with 128 of these already granted. Xiaomi’s self-developed electric drive also received the “Global New Energy Vehicle Frontier and Innovation Technology Award” during the 2024 World New Energy Vehicle Conference.

Regarding batteries, the Xiaomi SU7 Ultra employs the same track-optimized, high-capacity battery pack as the prototype, featuring the CATL Qilin 2.0 battery. This battery pack is among the most powerful mass-produced options available, capable of a maximum discharge rate of 16C and a maximum discharge power of 1330kW. Even with just 20% battery remaining, the discharge power still reaches 800kW.

The highest charging rate stands at 5.2C, allowing for charging from 10% to 80% to complete in only 11 minutes. With a CLTC range of 630km, it merges high performance with extended reach. Integrating Xiaomi’s CTB battery technology enhances volume efficiency, and the battery pack includes dual large-surface active cooling to ensure optimal heat dissipation and safety.

For electric vehicles to sustain high performance over time, they require exceptional cooling capabilities. The Xiaomi SU7 Ultra has seen upgrades in its cooling system across all components. The efficiencies of the compressor, water pump, cooling fan, and radiator have been significantly improved. The maximum heat dissipation per minute reaches 2.7*10^6J, which is three times greater than the heat dissipation capacity of a standard EV, effectively preventing overheating during two back-to-back laps on the Nürburgring Nordschleife.

The Xiaomi SU7 Ultra comes equipped with a top-tier braking system, featuring carbon ceramic brake discs, high-performance fixed calipers, and specialized brake pads. It can halt from 100km/h to 0 in just 30.8m, and it successfully completes ten consecutive braking tests from 180km/h to 0 without any fading. Furthermore, the Xiaomi SU7 Ultra boasts the largest racing-grade carbon ceramic brake discs available in a sports sedan.

The front brake disc has a diameter of 430mm, the largest size for a carbon ceramic disc on a sedan. In comparison to conventional steel brake discs, carbon ceramic discs can withstand higher temperatures, with a maximum operating threshold exceeding 1300°C, more than twice that of typical steel discs. They are also more resilient against wear and lighter in weight. Their lifespan exceeds 500,000km while reducing the vehicle’s total weight by 57kg.

Additionally, the Xiaomi SU7 Ultra is fitted with Akebono®️ high-performance brake calipers. The fixed calipers have six pistons at the front and four at the rear, with working areas of 148cm² and 93cm² respectively, offering outstanding braking power. The vehicle also features endurance racing-level ENDLESS®️ high-performance brake pads*, which can function at a maximum temperature of 1100°C while maintaining stable braking performance. Moreover, the efficient braking energy regeneration system can achieve a maximum deceleration of 0.6g and regeneration power exceeding 400kW, significantly easing the load on the braking system.

Exceptional driving experience: The track-tuned chassis system of the Xiaomi SU7 Ultra is specifically designed for the Nürburgring Nordschleife, making it ready for the track straight from the factory.

The Xiaomi SU7 Ultra is equipped with a peak chassis system fine-tuned for the Nürburgring Nordschleife, ensuring improved chassis control and an elevated handling capability. As a four-door “race car,” it is track-ready right from the factory.

For its suspension, the Xiaomi SU7 Ultra incorporates dual-chamber air springs and high-performance continuous damping control (CDC), which allow for a broader range of spring stiffness and damping adjustments, making the standard mode more comfortable and the sport mode more dynamic. It also offers track enthusiasts a professional-grade Bilstein®️ EVO T1 coilover shock absorber kit, featuring a maximum spring stiffness of 300N/mm and a peak damping force of 9000N.

With 10 levels for compression and rebound damping adjustment, it significantly enhances the control limits during acceleration, braking, and cornering, providing unmatched support during track cornering. Users have the option to choose between dual-chamber air springs or coilover shock absorbers based on their preferences.

Regarding chassis control, the Xiaomi SU7 Ultra’s tri-motor setup facilitates torque-vectoring control. The torque from the three electric motors is distributed independently and dynamically, with the torque adjusted 500 times per second, greatly improving the vehicle’s handling and stability. When cornering, the system directs more torque to the outer wheels to enhance steering response and speed during corner exits.

If one wheel encounters slippery conditions, the system can instantaneously adjust the torque to the left and right wheels within milliseconds to assist the vehicle in navigating more effectively. This torque-vectoring control system adheres to the highest ASIL-D functional safety standard in the industry, with a maximum fault-coordinated shutdown time of just 14ms, ensuring the vehicle’s safety.

The Xiaomi SU7 Ultra’s chassis system has experienced extensive tuning at the Nürburgring Nordschleife. Beginning June 11, the Xiaomi SU7 Ultra has undergone real-world testing at the Nürburgring Nordschleife, covering over 3,000 kilometers over six weeks of refinement. Utilizing the globally recognized Nürburgring Nordschleife as a testing ground, the Xiaomi SU7 Ultra aims to provide users with optimal handling performance.

The Xiaomi SU7 Ultra comes equipped with a tri-motor configuration, a high-capacity battery pack, an optimally designed cooling system for the track, a carbon ceramic braking system, dual-chamber air springs, and advanced continuous damping control, allowing you to experience the thrill of extreme speed on the track. Users with more demanding track requirements can select top-tier equipment such as Bilstein®️ coilover shock absorbers* and ENDLESS®️ high-performance brake pads*.

Unmatched immersive experience: The exclusive cockpit UI and sporty audio, along with a dedicated Racetrack Master app for competitive driving.

To enhance its outstanding performance and chassis system, the Xiaomi SU7 Ultra has introduced a novel visual and auditory interaction experience through an exclusive Racetrack Master app, delivering an unparalleled immersive experience.

The central control screen, instrument cluster, and various UI elements of the HUD have been upgraded in the Xiaomi SU7 Ultra, resulting in a cooler and sportier exclusive cockpit UI design. The Xiaomi SU7 Ultra features three new sound wave options for sports: super power, super sound, and super pulse, and includes a 40W external speaker to support these sound waves.

Xiaomi SU7 Ultra has developed a robust exclusive Racetrack Master app specifically for track users. This app can showcase all vehicle information, including real-time lap times, vehicle condition, and adjustments to driving modes. When drivers begin racing, they have the ability to set a benchmark lap time that shows the difference from the current lap time in real time, maximizing competitive engagement. Post-race, it provides historical lap statistics and visual analysis of results, along with key metrics like maximum speed, maximum G-force, steering wheel angle, and braking force. Drivers can easily export and share their race footage with just a click.

Currently, Xiaomi EV has partnered with 20 professional race tracks across China, adding track maps and rankings to the dedicated racetrack application. More professional tracks are expected to be incorporated in the future.

The Xiaomi SU7 Ultra presents a range of driving modes suitable for both track safety and everyday use. For track performance, it offers several options including endurance mode, qualifying mode, drift mode, and master custom mode, providing endless driving excitement. For daily driving needs, available modes include beginner mode, economy mode, wet mode, sport mode, and custom mode, among others. To ensure safe operation, drivers must possess driving skills or certification to activate track mode for the first time in the Xiaomi SU7 Ultra. The daily driving modes come with limitations on power and speed.

The Xiaomi SU7 Ultra is a performance powerhouse that resonates with individuals who seek the excitement of driving, aspire for the remarkable, and aim to make the most of each day. It is ideally suited for both everyday driving and racetrack adventures. This luxury tech sedan radiates a robust sense of sophistication and technology, making it appropriate for daily use, whether for business travel or meetings. It boasts top-tier technological features and provides unaltered performance capabilities for high-speed thrills.

Pre-orders for the Xiaomi SU7 Ultra are now open, with the official launch set for March 2025. The pre-order price is established at 814,900 yuan, requiring a deposit of 10,000 yuan to express intent. Those who pre-order will enjoy priority delivery.

The Xiaomi SU7 has made its initial official debut in Malaysia

The Malaysian unveiling of the Xiaomi SU7 took place at Suria KLCC, marking the first look at the company’s inaugural electric vehicle.

However, don’t rush to Suria KLCC in hopes of placing an order for the SU7. Xiaomi has not yet started accepting orders for this EV in Malaysia.

In reality, the launch of the SU7 in this market is not imminent. Xiaomi brought the EV to Malaysia primarily for a technology showcase, akin to Tesla’s approach with the Cybertruck.

The SU7 is featured as part of the Human x Car x Home ecosystem display, which is included in the Xiaomi 14T Series roadshow currently ongoing at Suria KLCC until this Sunday, October 20. Attendees can examine the EV in person at the event, although those who purchase the Xiaomi 14T Series during the showcase will be granted access to a priority lane.

At its core, Xiaomi HyperOS drives the concept of the company’s Human x Car x Home ecosystem, focusing on deep collaboration across its range of smart devices, spanning personal gadgets and home automation to vehicles. This concept represents an evolution of the firm’s earlier Smartphone x AIoT philosophy.

It also underscores the significance of vehicles in Xiaomi’s broader vision, particularly as the company enters the EV market. When Xiaomi first unveiled the SU7 internationally at Mobile World Congress earlier this year, they stated that their ecosystem comprises over 200 product categories, involving 600 million devices globally, covering more than 95% of users’ daily needs.

A Comparison of BYD SEAL EV and Xiaomi SU7

In the rapidly expanding market for new energy vehicles, the question, “Which pure electric vehicle should young buyers consider first?” has gained notable traction. As two eagerly awaited pure electric sedans, the refreshed BYD SEAL EV and Xiaomi SU7 each captivate young consumers with their distinct appeal. Today, we’ll conduct a straightforward comparison based on exterior design, interior features, power, and smart capabilities—key aspects for young buyers—to determine which vehicle better satisfies their preferences.

Exterior Design: Distinctive Styles, Each with Its Own Charm

What do young consumers prioritize when purchasing a vehicle? Naturally, the exterior comes first! The BYD SEAL EV showcases the brand’s hallmark minimalist design, characterized by sleek lines and a closed front grille that underscores its identity as an electric vehicle. Its fastback silhouette, paired with concealed door handles, creates a robust sporty aesthetic. Overall, the BYD SEAL EV’s exterior design adheres to contemporary aesthetic trends while exuding energy and style.

Conversely, the Xiaomi SU7 adopts a subtle sporty design, with its body profile reminiscent of a Porsche Taycan. The vehicle features a notably low design, incorporating semi-concealed door handles and a streamlined body that produces a visually striking appearance. Additionally, the Xiaomi SU7 is equipped with lidar and an electric spoiler, enhancing its uniqueness and technological allure, marking features that are uncommon among traditional brand offerings.

Interior layout: A combination of comfort and technology

Inside, the BYD SEAL EV prioritizes functionality and a sense of advanced technology. From a design standpoint, the new BYD SEAL EV has received a substantial interior makeover, introducing practical elements such as a knee airbag for front passengers, a central airbag for the front row, and soundproof glass for the rear seats, significantly boosting passenger comfort. Furthermore, the BYD SEAL EV includes four-zone voice recognition and 50W wireless fast charging, fulfilling young consumers’ desire for intelligent connectivity.

The Xiaomi SU7’s interior design features a minimalist and sporty aesthetic, emphasizing sustainable materials and high quality. A standout aspect of the central console is the large 16.1-inch display, which delivers exceptional display quality along with top-notch smoothness and functionality. The Xiaomi SU7’s smart cockpit system utilizes HyperOS, creating a deep integration with Xiaomi smartphones that results in an outstanding smart user experience. For fans of the Xiaomi ecosystem, this is truly worth considering.

Power Performance: A Battle of Strength and Efficiency

The BYD SEAL EV has received considerable upgrades in power, beginning with a 231-horsepower rear-wheel-drive motor, while the high-power variant achieves 313 horsepower. Although the driving range has slightly decreased in some models to boost power, the SEAL EV’s overall power performance remains robust. Additionally, the incorporation of the 800V high-voltage platform has significantly enhanced charging speeds, improving the overall user experience.

The Xiaomi SU7 also showcases remarkable power performance. The standard version of this vehicle can accelerate from 0 to 100 km/h in just 5.28 seconds, rivaling entry-level sports cars. In comparison to the BYD SEAL EV, this car’s performance is quite impressive. Furthermore, regarding chassis tuning, the Xiaomi SU7 excels in sporty performance and comfort. It effectively minimizes minor vibrations and offers superb support during cornering, making it feel particularly confident on mountain roads.

Smart Features: A Dual Assurance of Technology and Safety

Now, let’s evaluate the intelligent functionalities. The BYD SEAL EV has received a thorough upgrade in its smart driving system, starting with a monocular perception camera. The smart driving variant features a mix of binocular cameras and lidar, facilitating intelligent driving capabilities both on highways and in urban settings. While it still trails behind leading manufacturers like Huawei, it sufficiently meets the requirements of most consumers.

The intelligent cockpit system of the Xiaomi SU7 also stands out. Its in-car interaction system, based on HyperOS, is a leader in terms of both smoothness and functionality. Additionally, the Xiaomi SU7 provides a variety of smart connectivity options, including direct integration with Xiaomi smartphones and voice control through Xiao Ai (the AI assistant). Users can even manage smart home devices with ease, offering an unparalleled smart experience.

Conclusion:

To summarize, both the BYD SEAL EV and the Xiaomi SU7 possess unique advantages, making it challenging to declare a definitive winner. The BYD SEAL EV has attracted considerable consumer interest thanks to its strong performance, extensive smart driving features, and comfortable seating experience. Conversely, the Xiaomi SU7 appeals to younger buyers with its sleek sporty design, exceptional chassis tuning, and advanced smart cockpit system. Ultimately, the choice between these two models lies in personal preferences and requirements. Whether opting for the BYD SEAL EV or the Xiaomi SU7, each stands out as an impressive electric sedan worth considering for younger consumers.

In a striking development influenced by the evolving business landscape and significant geopolitical factors, the Chinese smartphone and consumer electronics powerhouse Xiaomi has entered the automotive industry. This strategic shift was extensively detailed by Lei Jun, the dynamic founder, chairman, and CEO of Xiaomi, during his speech for the year 2024.

The Trigger for Transformation

This monumental transition was signaled by a significant journey that commenced on January 15, 2021. Xiaomi encountered an unexpected crisis when the United States imposed sanctions against the company. Lei Jun described learning about the sanctions as “a bolt from the blue,” which severely impacted operations and future prospects.

During one of the most crucial board meetings, the question was posed: “If mobile phones can no longer be produced, what will happen to our 30,000 to 40,000 employees?” This marked the beginning of a determined effort to seek alternative paths and create new opportunities, eventually leading Xiaomi to consider entering the automotive sector.

Encouragement and Assistance from Industry Leaders
In that moment of uncertainty, Lei Jun was not alone. Industry colleagues, including Li Bin from NIO and He Xiaopeng from Xpeng Motors, offered him strong encouragement during this challenging time. These pioneers in China’s electric vehicle market were well aware of the challenges and possibilities in vehicle manufacturing and encouraged Xiaomi to pursue this direction.

The Choice to Manufacture Vehicles

Shortly thereafter, Lei Jun embarked on his next significant business venture to enter the car manufacturing sector. On March 20, 2021, Xiaomi made an official announcement regarding its entry into the automobile manufacturing industry, which became Lei Jun’s final entrepreneurial undertaking. The response was overwhelming; in just over three years, Xiaomi Auto garnered approximately 380,000 resumes, reflecting the public’s strong interest and confidence in this new path for Xiaomi.

Strategic Intent of Xiaomi

For Xiaomi, launching a car manufacturing business is more than just diversification—it’s a strategic initiative aimed at leveraging its technological expertise and consumer electronics strengths in the rapidly expanding electric vehicle market. Xiaomi plans to enhance its innovative capabilities in areas such as AI, IoT, and software in conjunction with automotive technology to create intelligent vehicles that deliver seamless user experiences.

Obstacles and Prospects Ahead

While the excitement and backing are palpable, Xiaomi’s entry into the automotive sector comes with its own set of challenges. The company will need to tackle:
- Technological Integration: Smoothly incorporating existing technologies into new automotive uses.
- Market Competition: Intense rivalry from established automotive brands and tech companies venturing into electric vehicles.
- Regulatory Landscape: Adhering to global automotive regulations and standards, which differ from those in the consumer electronics field.
- Lei Jun’s address at the Mi Mixer event marked a definitive turning point for Xiaomi. For the company, stepping into the car manufacturing industry symbolizes more than mere diversification; it represents a transformation in its identity and strategy, influenced by external pressures and internal strengths. The global audience will be keenly watching how Xiaomi accelerates this vision with innovative contributions to the automotive industry.
Xiaomi has received a license from the Chinese government to produce electric vehicles, setting the stage for its third model.

On July 12, the Chinese authorities announced that Xiaomi could independently assemble electric vehicles, meaning the smartphone company has passed the necessary regulations to increase production without relying on its usual car manufacturing partner, BAIC.

This development is significant as the approval from Chinese regulators will facilitate a smooth scale-up in production for Xiaomi, which has increased its delivery goal for the current year to 120,000 units from 72,000 and aims to attract a broader customer base with future models.

According to the public registration filings released by China’s Ministry of Industry and Information Technology (MIIT) on July 12, Xiaomi is now recognized as one of the “all-electric passenger car manufacturers.”

The company has also revised its registration filing for its initial consumer vehicle, the SU7, with the country’s main industry regulatory authority, now featuring the “Xiaomi” branding on the back instead of “Beijing Xiaomi” as was previously shown in the MIIT images.

The well-received sedan has consistently been produced at Xiaomi’s facility in the Beijing Economic and Technological Development Zone, but its production application was initially submitted under the name of a BAIC subsidiary, TechNode reported.

Xiaomi awaited final clearance from MIIT after receiving initial approval from China’s state planning authority for EV manufacturing, as reported by Reuters last August, while also seeking a partner for its second vehicle production. Its plant in Beijing has an annual production capacity of 150,000 cars.

For context, Xiaomi reached the milestone of 10,000 units sold by June, marking its third month of deliveries, contributing to a year-to-date volume of nearly 26,000 units for its competitor to Tesla’s Model 3.

In May, President Lu Weibing expressed to investors that the target for car deliveries this year is 120,000, which exceeds the initial goal of 72,000 units disclosed by CEO Lei Jun during Xiaomi’s annual investor conference in April, as reported by CNBC.

Starting from RMB 215,900 (equivalent to $29,881), the stylish all-electric Xiaomi SU7, resembling the Porsche Taycan, has achieved notable success in China, amassing 88,898 pre-orders within 24 hours of its launch on March 28. The company has faced pressure to ensure timely deliveries since then.

Additionally, Xiaomi is racing to launch its second vehicle, an all-electric SUV, in the first half of next year, while plans indicate that the third model may be an extended-range hybrid (EREV) aimed at Chinese families, anticipated for release in 2026.

Xiaomi’s impressive growth in 2024 is evident in its interim financial report for the first half of the year, showcasing significant success and rapid expansion across its main business areas, reinforcing its global position in smartphones, AIoT, and smart devices. With strong financial outcomes, record-setting shipments, and ambitious growth strategies, Xiaomi is gearing up for a prominent market presence for the rest of the year.

In financial highlights, Xiaomi’s operating income over the first six months of 2024 increased to 164.395 billion yuan, reflecting a substantial year-on-year growth of 29.62%. This strong revenue increase was accompanied by a 17.86% growth in net profit, reaching a total of 9.28 billion yuan. These results demonstrate Xiaomi’s capacity to leverage its diverse product offerings and global growth.

The mobile phone and AIoT sectors remain crucial for Xiaomi, generating 158 billion yuan in revenue—an increase of 24.6% compared to 2023. Notably, smartphone shipments achieved an impressive total of 82.2 million units, significantly boosting the company’s overall financial success. Revenue from smartphones rose by 29.9%, from 71.6 billion yuan in the same period last year to an outstanding 93 billion yuan for the first half of 2024.

Xiaomi’s IoT and consumer product sector, encompassing smart home devices and wearables, also experienced robust growth. This segment’s revenue increased by 20.6% year-over-year, totaling 47.1 billion yuan, with the number of connected IoT devices exceeding 822.2 million, reflecting a 25.6% rise compared to last year.

In addition to its established strengths in mobile and IoT, Xiaomi has seen advancements in its innovative pursuits, including smart electric vehicles. These segments generated 6.4 billion yuan in revenue, highlighting Xiaomi’s strategic investments in emerging fields. The company’s entry into the electric vehicle market has yielded the delivery of 27,367 units within just six months, underscoring its effective diversification strategy.

Wearables also performed exceptionally well in the first half of the year, with revenue soaring by 37.4% year-on-year, mostly driven by the growing popularity of smartwatches and TWS headphones in global markets.

Xiaomi’s smart home devices experienced notable growth of 40.5% year-on-year. This increase was particularly linked to a surge in shipments of air conditioners, refrigerators, and washing machines within mainland China.

Global Expansion and User Growth

Xiaomi is extending its reach, both in terms of user base and physical retail locations. The company now boasts 675.8 million global monthly active users, reflecting an 11.5% year-on-year rise. Xiaomi’s shift toward innovative retail strategies has resulted in over 12,000 physical stores in mainland China as of June 30, 2024.

Additionally, Xiaomi’s revenue from international markets amounted to 75.9 billion yuan, which constitutes 46.2% of its total revenue, underscoring the brand’s robust international presence. Its ventures into various regions have proven successful, establishing Xiaomi as a leading player in consumer electronics globally.

Conclusion: A Pivotal Year of Innovation

2024 is turning out to be a significant year for Xiaomi Group, marked by dramatic revenue growth, innovative product releases, and entry into new sectors such as electric vehicles. The company’s mobile and AIoT divisions continue to excel, bolstered by strategic investments in innovation and retail. As Xiaomi broadens its global ecosystem, it’s clear that the company’s ambition of creating a connected world of smart devices is nearing realization. The next six months will be crucial as Xiaomi ventures into new areas and strengthens its presence in key global markets.

On Monday, Xiaomi Corp announced its goal to deliver 130,000 electric vehicles this year, increasing its forecast for the third time due to a 30.5% rise in third-quarter revenue.

CEO Lei Jun revealed via social media that the electronics giant was updating its target to deliver 120,000 units of its first EV, the SU7 sedan, as demand continues to climb. This new goal significantly surpasses the initial aim of 76,000 set at the launch of the SU7 earlier this year.

The company introduced the car in March, taking design inspiration from Porsche, and entered the competitive Chinese EV market with an appealing price point of under $30,000 for the base version, which is $4,000 less than Tesla’s Model 3 in China.

Sales of electric and plug-in hybrid vehicles in China have now accounted for over half of total sales in the largest auto market globally. In October, this category saw a 56.7% increase compared to the previous year, marking the fourth consecutive month battery-powered vehicles, including plug-ins, outperformed gasoline cars.

To meet the rising demand, Xiaomi has increased factory production shifts since June and has launched the premium SU7 Ultra model which is priced over $110,000.

Following the earnings call, Xiaomi’s President Lu Weibing mentioned that their factory currently has the capability to produce 20,000 cars per month, with potential for further expansion.

“Our investment remains significant as we continue enhancing our hardware and software. The ultimate delivery numbers are not the main focus; we are maintaining strong investments and prioritizing R&D for new models,” he stated.

One area of focus for Xiaomi involves the development of autonomous driving technology.

AUTO BUSINESS STILL OPERATING AT A LOSS

Revenue for the quarter ending September 30 was reported at 92.5 billion yuan ($12.77 billion), surpassing an LSEG consensus estimate from 15 analysts which projected 91.1 billion yuan.

According to Huatai Securities, Xiaomi is expected to deliver 400,000 electric vehicles by 2025, at which point electric cars may contribute approximately one-fifth of revenue, compared to the current 8%.

However, Xiaomi’s automotive division is still facing losses. This segment reported an adjusted loss of 1.5 billion yuan for the quarter with a gross profit margin of 17.1%.

During this quarter, Xiaomi retained its status as the world’s third-largest smartphone manufacturer, with shipments reaching 42.8 million units, a 3% increase, capturing 14% of the market as reported by research firm Canalys.

Lu indicated that the company intends to grow its offline retail outlet count in mainland China from 13,000 to 15,000 by year-end, and aims for 20,000 by next year, while making substantial investments in technology to increase market share.

Xiaomi reported a 4.4% rise in adjusted net profit to 6.25 billion yuan, exceeding a consensus estimate of 5.92 billion yuan.
November 18, 2024
Virtual Reality (VR) has had a major impact on the healthcare sector
Virtual Reality (VR) has significantly influenced the healthcare industry, with extensive applications in both training and patient education and care.

In the realm of psychology specifically, numerous studies indicate that VR exposure therapy can provide various benefits for patients.

But what precisely does VR exposure therapy entail, and how is it implemented?

What is VR Exposure Therapy?

To comprehend what Virtual Reality Exposure Therapy (VRET) is, we first need to break down the term into its two parts: VR and exposure therapy.

Virtual Reality (VR)

VR refers to computer-generated technology designed to create a simulated environment. Utilizing VR headsets, a user finds themselves immersed in a particular 3D world that they can engage with.

Within this environment, many senses can be activated (sight, sound, touch, and occasionally smell), which helps to fully immerse the user in the artificial setting.

Exposure Therapy

The American Psychological Association defines exposure therapy as a psychological treatment developed to assist individuals in confronting their fears.

Individuals with fears (such as heights, flying, or spiders) often avoid situations or activities that trigger these fears. Although this may temporarily alleviate anxiety, over time, the fear may intensify.

In exposure therapy, a psychologist establishes a secure environment that incorporates those fears, enabling the patient to confront them and ultimately lessen avoidance and anxiety.

VRET

Now that we grasp the individual meanings of VR and exposure therapy, we can understand that VRET is a form of exposure therapy that employs VR technology to help expose patients to a safe environment where they can face and diminish their fears.

How do patients begin with VRET?

Typically, patients start VRET by getting acquainted with their therapist and discussing in detail what led to their trauma. The therapist will then create a customized VRET environment designed specifically for that patient.

To commence the therapy, patients will don a VR headset featuring a simulated environment that replicates their trauma. There may be sounds, sights, vibrations, or smells aimed at re-creating the experience and eliciting an emotional reaction.

This process exposes patients to what they may be trying to evade. Through this method, they can confront the situations that instill the most fear in them.

Following the session, patients engage in a discussion about their experience with their therapist, who will gain insight into their trigger points and how to best assist them in recovery as they move forward.

Is it effective?

VRET is still a relatively recent form of therapy, so research continues regarding the complete advantages of this approach. Nonetheless, several studies have already demonstrated its potential effectiveness.

For instance, Jimmy Castellanos, a Veteran Marine Corps member, experienced post-traumatic stress disorder for an extended period after his service in Iraq. His psychiatrist suggested this method, which virtually transported him back to the traumatic memory repeatedly until his triggers ceased to generate anxiety.

Castellanos remarked on the experience:
- It was an entirely different experience. I don’t recall having a physical reaction…In just 13 weeks, I had transformed who I had been for the prior ten years. Before the treatment, 80-90 percent of my dreams were related to Iraq. Now, I can’t remember the last time I had one. I lead a completely different life now.
- It is well known that soldiers returning from combat zones frequently suffer from PTSD. Recent advancements in VR technology have enabled these veterans (and many patients dealing with PTSD, anxiety, etc.) to finally receive the assistance they need and deserve.
Why is it effective?

VRET has demonstrated remarkable benefits in addressing various disorders, particularly PTSD, anxiety, and phobias.

Regrettably, these disorders are currently on the rise — The CDC has reported a notable increase in anxiety disorder symptoms. Many healthcare professionals in the U.S. have also noted rising PTSD rates amid the ongoing COVID-19 pandemic.

The effectiveness of VRET lies in the fact that individuals often develop anxiety and avoid situations that may remind them of a traumatic event. However, VRET encourages them to confront such situations.

For example, a war veteran with PTSD triggered by military combat might react strongly to the sound of fireworks. VRET allows them to face such triggers in a controlled setting. With prolonged exposure, they learn coping mechanisms for the anxiety and reframe their thoughts and feelings regarding a particular event.

Ultimately, the patient becomes accustomed to the triggers, embraces the experience, and their anxiety or stress response diminishes in intensity.

The same principle applies to using VRET for other forms of PTSD, anxiety, and phobias.

VRET vs. In Vivo Therapy

Historically, many psychologists have utilized “in vivo” therapy, which involves guiding patients through exposure-based activities face-to-face. For instance, an individual dealing with agoraphobia may be taken to a public location to help confront their fears, or someone who has a fear of flying might visit the airport with their therapist to replicate the experience of boarding a plane.

While this therapy can be effective, it often depends on the patient’s ability to move and their access to environments that allow them to face their fears without becoming overwhelmed, making it challenging to identify appropriate settings. In contrast, VRET enables therapists to adjust the exposure intensity according to the patient’s circumstances, supporting them in gradually confronting their fears over time. Furthermore, this type of therapy can occur remotely from any location, broadening access to patients beyond those who may benefit from in vivo therapy.

VRET represents a promising new approach for various mental health challenges, and this groundbreaking technology is likely to positively influence the lives of many patients.

Augmented reality, or AR, involves overlaying digital information onto real physical objects or environments—think of the Pokemon Go phenomenon from a few years ago.

However, AR can serve more purposes than just capturing animated characters in real life. In the medical field, it has applications in training, educating patients, assisting during surgeries, and other related functions.

Advances in both hardware and software have paved the way for numerous new and cost-effective applications of AR in healthcare, even for small medical practices. AR tools can be utilized through specific headsets, special eyewear like Google Glass, smartphones or tablets, or other specialized AR devices.

Let’s explore some of the ways AR can enhance healthcare in medical education and practice, as well as in consumer health applications.

AR for Medical Students and Practitioners

For medical professionals—including surgeons and nurses—understanding a patient’s anatomy is crucial before performing any medical procedure. AR can assist by showing visualizations of what lies beneath the skin, increasing accuracy in injections or incisions, or simply providing a clearer view of human anatomy.

At Case Western University’s medical school, for example, students can utilize Microsoft’s Hololens to observe a large 3D model of the human body, allowing them to navigate holographic representations of various tissue layers, muscle, and bone, thus enhancing their anatomical knowledge beyond what’s available during real procedures.

Nurses can employ AR handheld scanner technology to visualize their patients’ veins, making it easier to locate the vein for blood draws or vaccinations. This innovation saves time in vein identification while ensuring patient comfort throughout the process.

Surgeons can leverage AR to gain an augmented sense of vision. By using AR headsets featuring eye displays that project images of the patient’s internal anatomy derived from CT scans, they can obtain a much clearer understanding of the underlying structures. At Johns Hopkins, neurosurgeons have implemented this technology for tasks such as inserting screws during spinal fusion and excising cancerous tumors from patients’ spines.

“When augmented reality is used in the operating room, it functions like having a GPS navigator before your eyes in an intuitive manner, eliminating the need to glance at a separate display to check the patient’s CT scan,” explains Timothy Witham, M.D., director of the Johns Hopkins Neurosurgery Spinal Fusion Laboratory and a professor of neurosurgery at the Johns Hopkins University School of Medicine.

AR has also proven advantageous in allowing doctors to minimize their teams to limit potential COVID-19 exposure. Imperial College Healthcare NHS Trust used Microsoft HoloLens 2 and Dynamics 365 Remote Assist to provide doctors with hands-free video consultations with other specialists during procedures while enabling them to access medical notes and X-rays within their direct line of sight.

“This means that all the information and specialist care you require at the patient’s bedside is readily available, all through one headset,” stated Dr. James Kinross, a surgeon and senior lecturer at Imperial College.

Although the technology is still in its early stages, it is likely that medical schools and practices worldwide will harness the advantages of AR-assisted surgeries and other medical procedures. AR provides access to in-depth insights into patient medical data and anatomical information, which medical professionals and students can learn from and rely on during their operations. It represents a cost-effective and convenient means of gaining “x-ray vision” to enhance their performance in their respective tasks.

AR for patient care and education
AR offers numerous advantages for patients as well

For instance, during a Google Glass trial with the Australian Breastfeeding Association, nursing mothers wore Google Glass while consulting with virtual lactation consultants. This allowed the consultants to view exactly what the mothers were experiencing and provide valuable feedback and advice to enhance their nursing sessions—without requiring the mothers to set their babies down.

Patients can enhance their understanding of drug interactions through an app that scans drugstore shelves and generates alerts about which over-the-counter medications might negatively interact with their existing prescription drugs. The app, offered by Cigna, aids patients in reducing anxiety, preventing adverse reactions, and boosting the efficacy of their medications.

Lastly, AR can encourage individuals to prioritize their health by gamifying physical activity. Applications like Pokemon Go and Ghostbusters World offer a fun and engaging experience that promotes increased walking or running as part of daily routines.

The augmented reality market in healthcare is currently expanding at a compound annual growth rate of 32.9%. As hardware and application developers create more affordable AR technologies, and as medical practitioners, educators, and consumers recognize their value, AR is expected to play a larger role in our healthcare experiences, both at home and during office or operating room visits.

Researchers are continually investigating and improving virtual reality exposure therapy for mental health care, specifically in treating PTSD, anxiety, OCD, and various other mental health issues.

Virtual reality is becoming an increasingly important tool across different industries, and healthcare is no exception. The introduction of virtual reality in surgical training, pain management, management of neurological disorders, pediatrics, and mental health care has yielded distinct and multifaceted advantages for the healthcare system. As research into virtual reality in healthcare progresses, an increasing number of mental health professionals are considering its role in exposure therapy for treating fear-based or anxiety-related conditions.

Exposure therapy, which dates back to the 1900s, has been extensively utilized to manage and treat mental health disorders. Despite its proven effectiveness, there are many limitations in terms of accessing and initiating exposure therapy, as well as maintaining its continuity. With personal and safety obstacles posing challenges, researchers have turned to innovative virtual reality technologies to address some of these issues. However, even the most advanced technologies face their own set of challenges.

Exposure Therapy

Exposure therapy is utilized to treat a variety of mental health disorders, including phobias, panic disorders, social anxiety, generalized anxiety, obsessive–compulsive disorder (OCD), and PTSD.

The American Psychological Association (APA) identifies four primary categories of exposure therapy: in vivo exposure, imaginal exposure, interoceptive exposure, and virtual reality exposure.

In vivo exposure involves real-life, direct interaction with a specific situation or activity that may trigger symptoms. For example, someone with acrophobia, or fear of heights, might go on a rollercoaster or ascend a mountain. A person afraid of public speaking may be asked to deliver a speech.

Imaginal exposure requires patients to vividly visualize the situation or object that provokes a fear response. For instance, someone with claustrophobia may need to imagine being in a confined space.

Interoceptive exposure therapy involves exposing the patient to a harmless physical sensation to help them understand that this feeling does not signify danger.

Overall, exposure therapy can be a complex treatment avenue because, even with appropriate diagnosis and support, fear can lead patients to hesitate or avoid treatment.

In addition to the challenges associated with starting treatment, realistic exposure therapy may not always be feasible. For instance, in vivo exposure could be dangerous or inaccessible, while the traditional alternative, imaginal exposure, can be difficult to regulate.

Virtual Reality Exposure Therapy

Perhaps the most innovative type of exposure therapy is virtual reality exposure therapy (VRET). VRET leverages VR technology to create an entirely virtual, immersive experience, providing exposure therapy in the comfort of an office, home, or healthcare facility.

According to the APA, virtual reality therapy is “a form of in vivo exposure therapy in which clients are active participants in a three-dimensional computer-generated interactive environment that allows them a sense of actual involvement in scenarios related to their presenting problems.”

VRET is a form of exposure therapy that utilizes computers to create virtual settings, which users experience through virtual reality headsets or head-mounted displays (HMDs). Patients may find VRET to be more manageable and appealing compared to other types of exposure therapy.

An article in Campbell Systematic Reviews states, “The primary goal of VR is to substitute sensory experiences from the actual world and establish a sense of presence for the user in the virtual realm. To engage with the user in real-time, the VR system gathers data about the user’s position and head movements using sensors and input tools like a head tracking system or joystick.”

Psychological Assessments

According to research published in Dialogues in Clinical Neuroscience, virtual reality technology opens up a unique opportunity for deeper mental health evaluations by immersing patients in real-life scenarios.

While traditional psychological assessments have evolved since their inception, they fall short in accurately reflecting the daily experiences of patients. Limitations in live exposure lead to difficulties in precisely evaluating anxiety, PTSD, phobias, and other mental health issues, potentially affecting recommended treatment strategies.

Using VR assessments, mental health practitioners can glean insights into a patient’s psychiatric condition through virtual exposure. For instance, a study featured in the Annals of General Psychiatry found that both real and virtual images of food triggered similar reactions in people with eating disorders, indicating that VR might offer reliable assessments across various mental health issues.

Phobias

The initial demonstration of VRET’s effectiveness was reported by Barbara Rothbaum, PhD, in the American Journal of Psychiatry, observed to aid in overcoming a fear of heights.

A study in Cognitive Behavior Therapy examining VRET for individuals with public speaking anxiety concluded that VRET can be an essential therapeutic tool when implemented correctly and as part of the suitable care regimen.

Anxiety

As noted in Campbell Systematic Reviews, VRET has been investigated as an adjunctive therapy combined with traditional cognitive behavioral therapy for individuals with social anxiety disorder (SAD).

Research has further compared VRET as a standalone mental health intervention for anxiety, highlighting similar outcomes between VR therapy and conventional treatments.

Cognitive behavioral therapy (CBT) is the standard treatment for SAD and addresses other prevalent comorbidities such as depression. According to the systematic review and meta-analysis featured in Campbell, many individuals with SAD delay or avoid treatment due to high costs, extensive travel requirements, and other obstacles.

Post-Traumatic Stress Disorder

An article from the University of Central Florida (UCF) discusses the potential of VR exposure therapy in addressing post-traumatic stress disorder (PTSD). In addition to facilitating straightforward assessments, a publication by Albert Rizzo in the Annals of the New York Academy of Sciences examined the capacity of VRET to replicate combat experiences for soldiers who served in Iraq or Afghanistan.

Rizzo emphasizes that a customized approach to exposure therapy for combat-related PTSD could yield more patient information and assist in tailoring treatments to their specific experiences, enabling healthcare providers to evaluate PTSD symptoms and adjust medications or therapeutic strategies as needed.

Despite evidence from individual studies showing the efficacy of VR technology, a systematic review and meta-analysis published in the International Journal of Environmental Research and Public Health does not indicate a robust link between VRET and PTSD.

Findings suggest that gradually increasing stimuli throughout a session, rather than in response to the patient, diminishes the effectiveness of virtual reality exposure therapy for PTSD patients.

“Unfortunately, standard VRET involves increasing the intensity and frequency of trauma-related stimuli as the session continues, rather than tailoring this to the subject’s reactions. This approach may hinder full immersion for PTSD patients, as trauma-related stimuli are not presented with respect to their responses,” researchers noted in the article.

An alternative to the conventional progression of VRET is VR-based graded exposure therapy (VR-GET). VR-GET is a revised version of VRET treatment that observes a patient’s reactions during PTSD therapy. By assessing a patient’s physiological and emotional responses, the therapist can adjust treatment protocols accordingly.

Limitations

Although numerous potential benefits of VRET are recognized, there are also notable limitations. One significant barrier is the cost of VR technology, which can be prohibitive for patients, clinicians, or healthcare systems lacking the financial means for high-tech headsets and other necessary VR components.

The use of virtual reality (VR) in clinical practice faces limitations beyond financial costs for researchers and clinicians. The Dialogues in Clinical Neuroscience states, “The biggest challenge to implementing VR in clinical settings currently is the scarcity of evidence-based VR programs that can be easily purchased and utilized by clinicians and researchers. Several laboratories globally are creating their own software and conducting tests, but these solutions are not yet available for public purchase. The limited commercially available products developed by software firms have not been evaluated to determine their safety and effectiveness.”

Furthermore, these virtual solutions present challenges regarding the time, funding, and resources necessary for the upkeep and enhancement of software and hardware. The possibility of technological failures also poses a significant risk for patients with delicate conditions, like those with panic disorders.

Additionally, ethical issues relating to data security, privacy, confidentiality, and technological hurdles exist.

While VR is not a complete substitute for psychotherapy or other psychiatric treatments, it may offer an innovative approach to various conditions; however, further research, development, and incorporation into the healthcare system are vital for ensuring safe, effective, and affordable care.

VR therapy employs a computer-generated environment as a treatment tool. An individual might use it to practice skills, confront fears in a secure setting, or build confidence in social situations.

VR therapy is not intended to replace traditional treatments for mental health disorders. Instead, it is often utilized by clinicians as a supplementary intervention.

For instance, a therapist might integrate VR into cognitive behavioral therapy (CBT), enabling a client to practice new skills in a more controlled setting compared to real life.

Numerous studies indicate that virtual reality therapy can effectively manage various mental health issues, including anxiety, depression, post-traumatic stress disorder (PTSD), and phobias.

VR therapy leverages virtual environments and scenarios as therapeutic tools. A user may don a headset or utilize a device to immerse themselves in the virtual realm and engage with it.

VR technology allows for the simulation of a vast array of environments and situations, making it advantageous for creating scenarios that are difficult to replicate in the real world or that may be too intimidating or hazardous.

Therapists might employ VR to help clients navigate real-life challenges, revisit past experiences, or assist individuals in facing their fears in a controlled manner.

Researchers published the first study on VR therapy over 25 years ago, and as technology has advanced, interest in this therapeutic tool has grown.

How does VR therapy function?

VR therapy operates by providing individuals with the opportunity to enact, practice, or revisit situations in a safe environment. This approach may:
- teach skills
- alleviate fears
- enhance confidence
- assist in processing past events
By eliminating risks present in the real world, VR can render frightening experiences more manageable. For instance, someone with a phobia might not feel ready to confront it in reality, but engaging with a simulation in VR could help them gradually acclimate to the feared object and understand that it does not pose a threat.

In this manner, VR may facilitate a connection between therapy and real-life experiences.

What conditions could benefit from VR?

VR therapy was initially designed to address phobias, but over the years, therapists have experimented with it for a range of mental health disorders.

Phobias

Therapists can utilize VR for exposure therapy, which is a fundamental component of phobia treatment. This method involves slowly introducing a person to their fear in small, manageable increments, ensuring their consent.

While exposure therapy can occur without VR, it is sometimes challenging to achieve. For example, someone who fears flying cannot simply take a brief flight and progressively build their tolerance. Additionally, encountering fears like wild animals could expose individuals to danger.

VR broadens the opportunities for exposure therapy. A systematic review from 2022, which examined 18 studies, found that this method improved nearly all types of specific phobias addressed in the reviewed research, including animal phobias and fears related to blood or injections.

PTSD

Exposure therapy may also be beneficial for PTSD, but, as with phobias, controlled exposure to a traumatic event can be difficult and potentially unsafe.

Several studies indicate that VR therapy serves as an alternative. For example, a 2019 review and meta-analysis of nine prior studies compared the outcomes of VR exposure therapy with no treatment.

Compared to participants who received no therapy, those undergoing VR therapy reported a reduction in PTSD symptoms, with benefits persisting for at least three months following the conclusion of treatment.

Social and emotional skills

Individuals can practice a variety of social and emotional skills through VR therapy. For instance, they might rehearse addressing a conflict with their partner or request a raise from their supervisor. This enables them to safely experiment with new skills while under professional guidance.

Anxiety and depression

A review published in 2019 examined earlier studies and highlighted the potential utility of virtual reality (VR) in treating various aspects of anxiety and depression. It could:
- assist individuals in understanding mental health
- visualize cognitive-behavioral therapy (CBT) techniques
- promote self-compassion
- simulate therapies like gardening or animal-assisted therapy
- A scoping review from 2021 evaluated nine prior studies that integrated VR with CBT and concluded it could be beneficial for treating anxiety and depression.
What is the cost of VR therapy?

Within a therapist’s office, VR therapy typically costs about the same as conventional psychotherapy. Insurance may cover VR therapy in a clinician’s setting if the therapist is recognized by the insurance provider.

Some therapists provide clients with VR devices to use at home to complement their sessions, while certain companies offer home VR units for self-care. Clients might rent these devices weekly, depending on the type of device.

How can one find VR therapy?

To explore VR therapy, an individual needs to locate a licensed psychotherapist who has access to a VR device. Online search engines and directories for therapists may assist with this process.

VR therapy may be appropriate for individuals who:
- experience specific phobias or fears
- wish to practice particular skills
  are not prepared or able to confront certain situations in real life
- Numerous companies provide home VR therapy through an app, allowing individuals to progress at their own pace. However, this format is not equivalent to traditional psychotherapy and may lack some benefits.
When should one seek assistance?

Individuals should seek help when any mental health concern negatively impacts their relationships, quality of life, or well-being, especially if self-care has not alleviated their symptoms. This support could be accessed through a doctor or any qualified therapist, regardless of whether they offer VR therapy.

It is crucial to seek help if someone has thoughts of self-harm or suicide.

Virtual reality therapy utilizes virtual reality to recreate various scenarios. This technique can assist individuals in acquiring new skills and addressing their fears in a secure setting. Initially, therapists employed it to treat phobias, but it is now used for a wide array of conditions.

VR can create a controlled environment to tackle situations that might feel overwhelming or hazardous in the real world, which makes it valuable for exposure therapy. Nonetheless, as with any form of therapy, it is vital to receive care from a licensed and experienced provider.

What is the experience of virtual reality exposure therapy like?

You will spend time interacting with your therapist and discussing the events that led to your trauma. Following this, your therapist will establish the setting for your virtual reality exposure therapy (VRET). You may use a VR headset or enter a dimly lit room filled with screens that produce an immersive environment echoing what your trauma felt like. The experience can involve sights, sounds, smells, and vibrations to further replicate the traumatic event and emotional response. This setup aims to help you face the situations that induce fear and anxiety in a safe and monitored setting. You will review these immersive experiences with your therapist. Medications and coping skills training might also be integrated with your therapy.

Please note: VRET may induce dizziness or headaches, particularly for individuals with brain injuries.

What do patients say?

“It was an entirely different experience. I don’t recall having the physiological response… In just 13 weeks, I had completely transformed from who I had been for the last ten years. Before the treatment, 80-90% of my dreams were related to Iraq. Now I can’t even remember the last time I had one. I am living in a completely new way now.” – Jimmy Castellanos, Veteran, U.S. Marine Corps

“The layers … they just peel back and reveal your core. Initially, you resist, but eventually you break down and face it, and it’s truly incredible.” – Kevin Tergliafera, Veteran, Army National Guard

Why does virtual reality exposure therapy work?

When certain individuals undergo a traumatic event, they might react naturally, resulting in a heightened fear reaction to stimuli like sights, sounds, or other elements that trigger memories of that trauma. This can lead them to avoid circumstances that incorporate those triggers, such as the sound and sight of fireworks for someone with PTSD from military combat. By exposing oneself to these triggers in virtual reality, individuals can confront their fears in a controlled environment.

Similar to traditional prolonged exposure treatment, this practice enables one to learn coping mechanisms and reevaluate thoughts regarding the traumatic incident. Ultimately, this can lead to becoming increasingly desensitized to the triggers and coming to terms with the experience. Over time, the stress responses to these triggers can diminish significantly.

How substantial is the evidence?

Research indicates that VRET may be effective in alleviating PTSD symptoms. Numerous studies have demonstrated that VRET is associated with a reduction in symptom severity for both PTSD and depression, and that the effectiveness of symptom relief tends to increase with the number of VRET sessions attended. These improvements have also been shown to persist over time, as seen in 3-month and 6-month follow-up evaluations.

A randomized controlled trial confirmed these findings, indicating that patients undergoing VRET reported a decrease in symptoms related to PTSD, depression, and anger. This research concluded that VRET is most beneficial when combined with additional traditional treatment methods. Despite encouraging initial results, further research is required.

What are the characteristics of effective virtual reality exposure therapy?

Locate a licensed psychologist or another qualified therapist who has experience with prolonged exposure therapy, including VRET for PTSD and/or TBI. It is advantageous if they have experience dealing with your specific trauma source.

Advantages of VRET Compared to Traditional Exposure Therapy
There are various advantages to participating in virtual reality exposure therapy. Virtual reality acts as a link between a simulated stressful environment and the real world.

Here are some of the advantages of VRET therapy:
- It can be more cost-effective compared to real-life exposures: A virtual environment provides a budget-friendly and practical option, especially in situations where repeated real-life exposure may be too costly and dangerous, such as fears related to flying, heights, or wild animals.
- Participants experience a sense of control: VRET employs specialized equipment and devices that create a very realistic feeling experience; however, if it becomes too overwhelming, the session can be halted at any point.
- It can be beneficial for individuals lacking access to other treatment options: With the expansion of availability and quality of VRET, it may be possible to offer mental health care to patients who would otherwise have limited access to treatment.
- It often yields enduring results: Several studies indicate that VRET can effectively address anxiety, PTSD, and depression, with symptoms remaining low during follow-up appointments. Although these previous findings are promising, additional research is necessary.
Challenges & Obstacles to VRET Therapy

While VRET demonstrates effectiveness, it remains an emerging field with several challenges and obstacles to consider.

Here are some potential barriers to virtual reality exposure therapy you may encounter:
- There has been a gradual acceptance of VR technology as a viable therapy option: Clinicians often prefer face-to-face treatments. Even when professionals are trained in VR, they seldom utilize it, partly due to misconceptions about this exposure-based method.
- Access to and selection of VRET technology is restricted: VR software, equipment, and guidance on how to use them are not readily accessible to all therapists. Additionally, the broad array of materials can make it challenging for professionals to determine what is appropriate for themselves and their clients.
- There is insufficient training available for clinicians, making it difficult to find a licensed provider: Despite the increasing interest in VRET, there’s a lack of training opportunities for professionals wishing to incorporate it into their practice. Furthermore, they may need to refresh their training each time new software or products become available.
- More research is essential for understanding its effectiveness: Currently, VRET primarily addresses anxiety-related disorders such as PTSD and specific phobias. However, with the growing efficacy and popularity of VRET, more data is required to extend its benefits to other mental health conditions.
What Equipment Is Typically Utilized in Virtual Reality Therapy?

To partake in virtual reality exposure therapy, a provider employs programmed computers, immersive devices, and artificially created settings that replicate reality through a simulated experience. The individual undergoing virtual reality exposure therapy is fitted with a headset that grants access to the virtual environment. Contemporary virtual reality equipment has been adapted for smartphones, enabling the use of gyroscopes and motion sensors to monitor head, body, and hand positions while also tracking subjective units of distress.

The equipment used during a virtual reality exposure therapy session is generally supplied by the therapist, although some individuals seeking this treatment may purchase their own headsets or goggles either through their therapist or online, with prices ranging from less than $50 to several hundred dollars.

Virtual reality exposure therapy is suitable for various populations and ages. Besides adults, children and teenagers can be great candidates for innovative methods to develop healthy coping strategies. Research investigating the effectiveness of virtual reality exposure therapy for adolescents aged 13 to 16 who faced public speaking fears due to social anxiety showed positive outcomes in helping them manage their symptoms.

Beyond its common applications, virtual reality is also being explored for treating sleep-wake disorders, enhancing sports performance, and addressing stress and test anxiety.

Virtual reality has been widely employed as a prolonged exposure technique for treating PTSD in military personnel. Significant funding from military sources has facilitated numerous studies to assess the effectiveness of this method. With virtual reality exposure therapy, providers can create an immersive, 360-degree interactive computer-simulated environment.

A meta-analysis that reviewed 14 studies involving military populations with PTSD demonstrated the high efficacy of virtual reality exposure therapy. Additionally, this therapy has been utilized in treating PTSD among military personnel through a program called Virtual Iraq. In this approach, soldiers use a head-mounted display and a gamepad to navigate a simulated Iraqi environment while traveling in a Humvee.

Systematic exposure to these feared stimuli and settings aims to alleviate anxiety and traumatic stress symptoms. Initial findings among the first group treated revealed significant reductions in PTSD symptoms, with 75% of participants no longer meeting the DSM-V criteria for the diagnosis. Another investigation using the Virtual Iraq framework reported approximately a fifty percent reduction in PTSD symptoms among veterans diagnosed with the condition.

Virtual reality exposure therapy serves as a practical exposure therapy method for anxiety disorders, with strong empirical support. Research findings uphold the use of VRET to address anxiety and phobia symptoms. This technology-based treatment approach enables clients to cope with anxiety by facing their fears through gradual or repeated exposure.

The ultimate aim is to alter thought patterns, behaviors, and reactions that hinder daily functioning. The feared stimuli can differ from individual to individual, but may include living creatures, objects, situations, activities, thoughts, mental images, physical sensations, or experiences. Encountering these feared stimuli can lead to the extinction of the fear response, which is beneficial compared to other forms of exposure therapy (e.g., in vivo exposure), as VRET allows access to the most feared cues.

VRET is increasingly employed for specific phobias, such as an intense fear of animals. The virtual settings can replicate feared animals or insects like spiders, snakes, or roaches. Recently, it has also been adapted for treating fear of public speaking, fear of heights, addiction, bullying, claustrophobia, depression, and eating disorders.

What contributes to the success of virtual reality exposure therapy? Virtual reality uses visual and often physical simulations to evoke sensory responses similar to real-life experiences. By integrating VR with various techniques, VRET enables individuals to confront and receive feedback in a secure environment, facilitating processing and diminishing established responses. Moreover, it allows individuals to gradually approach fears that might be harder to tackle in the real world. Ultimately, VRET provides a safe exposure experience, encouraging individuals to challenge their reactions, process stimuli, and alleviate symptoms, ultimately enabling them to engage more in real-life activities.

The effectiveness of exposure therapy, especially when combining multiple theories, can be explained through habituation, extinction, emotional processing, and self-efficacy theories. Each theory or a combination thereof can elucidate the success of virtual reality exposure therapy:
- Habituation theory is utilized by repeatedly presenting a stimulus (for instance, an individual who has been attacked in a park and now avoids parks utilizes VR to be immersed in a virtual “park” environment) to lessen anxiety and boost familiarity.
- Extinction theory is employed to diminish the conditioned response (such as fear, avoidance, anxiety) by weakening the reinforcement of an unconditioned stimulus (the experience of being attacked in the park); this association is weakened through repeated exposure to a conditioned stimulus (the park) without the occurrence of the unconditioned stimulus.
- Emotional processing is implemented by repeatedly facing a stimulus (the park) to confront the response (panic, anxiety) and the unhealthy beliefs (for example, “I’m too weak to defend myself, I’m foolish for not anticipating that”) that were initially ingrained in one’s memory.
- Self-efficacy is fostered by acquiring techniques to handle or master a fear-inducing situation, which affects a fear or anxiety response; by mastering these skills, one acknowledges their ability to manage a frightening circumstance and can use this knowledge with similar stimuli.
What to Expect at Your First Session

Prior to commencing virtual reality exposure therapy, the therapist will conduct an initial consultation to evaluate if a prospective client is suitable before starting. Once an individual is considered eligible for VRET, the therapist will administer a biopsychosocial assessment during the first session, allowing them to gather comprehensive information about their client and their therapy objectives prior to entering the treatment phase.

After this assessment, the next step will involve treatment planning, which includes providing fundamental education about the specifics, establishing expectations, and allowing the client an opportunity to ask questions or voice any concerns. It’s important to note that each therapist may have different operating procedures, but they will typically include many of the standard processes discussed. The treatment process will then commence and can occur either in person or through virtual means.

What Is a Typical VRET Session Like?

During the exposure therapy session, where the therapist may address a specific phobia or trauma, the client will encounter exposure to feared stimuli or environments. By using VR equipment such as a headset or goggles, the client will have direct access to the simulated or artificial environment while gradually increasing the intensity of their exposure to the stimuli.

Thanks to the flexibility in regulating the simulated setting, the individual undergoing treatment should recognize that the therapist can lessen or eliminate exposure to the fear at any moment. With the inclusion of biofeedback equipment, physiological sensations can be monitored and recorded using sensors between treatments. If the session is being conducted virtually, all that is required are the headset or goggles and access to a smartphone since the virtual environment is navigated through an app on the individual’s phone, which the provider controls.

VRET Examples

While the fundamental procedures of VRET will be consistent for everyone, there may be some differences based on the client’s experiences and the specific VRET equipment being utilized.

Here are a few instances of VRET in practice:

VRET for a Phobia

Dan, a 31-year-old man, recently purchased a house and soon discovered it was infested with spiders. Dan suffers from a severe phobia of spiders, often resulting in panic attacks. He moved in with his parents after scheduling pest control for his new home but has struggled to go back and even see how the house looks due to his fear and anxiety surrounding the spiders.

Dan consults a VRET therapist to address this issue, as he wishes to return to his new residence. He and his therapist begin with an exposure hierarchy, listing his fears to evaluate comfortability. They decide on habituation and self-efficacy techniques.

In each session, he wears VR goggles and biofeedback sensors because of his panic attacks and observes virtual spiders approaching him. He engages in this for an extended duration in each session while his therapist monitors his sensors. Following each exposure, they discuss the emotions that arise, partake in educational activities and coping practice, and occasionally repeat the exposure. Eventually, Dan feels comfortable enough to visit his home and, after 2 months, succeeds in moving back in with ongoing therapy.

VRET for PTSD

Jane, a 29-year-old woman, has recently returned from her third year-long deployment in an active combat zone. Although she managed to adjust well after her first return, she has recently started experiencing symptoms of post-traumatic stress disorder (PTSD). During her debriefing, she frequently discussed her symptoms and continues to face flashbacks, hypervigilance, nightmares, irritable behavior, and challenges at work and with her family, even 6 months later as a result.

Jane began working with a VRET therapist knowledgeable in military matters to start addressing some of her symptoms, hoping to reclaim her life. Loud noises triggered her significantly, often causing reactions that others saw as exaggerated. Together, Jane and her therapist commenced sessions using Virtual Iraq VRET, employing a head-mounted display and game pad, gradually increasing the duration while identifying specific triggers and applying coping strategies and skills as they arose.

Over time, Jane managed to return to her job and developed an action plan with her employers and family for managing loud noises and particularly disruptive flashbacks.

VRET for Social Anxiety

Taylor, a 19-year-old, was frequently labeled as shy during high school. However, Taylor struggled significantly in crowded settings, never took part in talent shows or groups that required extensive social interaction, and often found socializing challenging. The school counselor suggested that Taylor might be experiencing social anxiety disorder (SAD), but did not facilitate any treatment.

After a year at college, Taylor continued to face many of the same challenges and found it difficult to connect with others, but desired friendships and connection while attending school. A therapist subsequently diagnosed Taylor with social anxiety following a comprehensive evaluation and recommended VRET. In therapy sessions, Taylor was engaged in social interactions and practiced communicating with several people for longer periods, using a headset and biofeedback sensors. Taylor was also encouraged to confront or challenge negative thoughts such as “I’m an idiot” that arose during interactions, both in real life and within the VR environment.

Eventually, Taylor formed a small circle of local friends as well as some online friends and started participating more actively in classes.

Who Can Provide VRET?

There is specialized training available that equips providers with the necessary skills for delivering virtual reality exposure therapy. Nevertheless, there are currently no specific certifications mandated for offering this type of treatment. Any mental health professional interested in this method can undergo training through organizations like Psious, which is a leading entity supplying VR equipment tailored for treating various mental health disorders.

Donna Davis, Ph.D., who directs the Oregon Reality Lab in Portland, Oregon, specializes in virtual reality therapy (VRT). She clarifies that VRT takes place in a computer-generated or 3-D environment and is entirely distinct from teletherapy. While teletherapy involves virtual talk therapy (like through Zoom), VRT focuses on utilizing a virtual setting, such as a computer game or headset. It is crucial to emphasize that for it to qualify as therapy, a licensed therapist must be present. Programs or videos aimed at relaxation or enhancing meditation do not qualify as VRT since there is no therapist involved.

A specific kind of VRT is known as virtual reality exposure therapy (VRET), which immerses individuals in a highly realistic 3-D environment. This is often accomplished using a headset, but not always. For instance, if someone has a fear of heights, the 3-D setting might feature a glass elevator to aid them in confronting their fear. VRET is also employed to support individuals with various phobias, as well as those experiencing post-traumatic stress disorder (PTSD) or victims of violence.

However, VRT does not always reach such an immersive level. Dr. Davis mentions that another version of VRT involves conversing with a therapist while assuming an avatar’s identity in a computer-generated setting. For example, Dr. Davis has been involved with a virtual reality support group for individuals with Parkinson’s disease on the online platform Second Life, where members create 3-D characters in an alternate universe. The group has been “meeting” consistently for over a decade. “Participants in the group develop an avatar, which allows them to feel more secure when opening up, as their actual physical identity remains hidden,” she states.

As VRT remains relatively new, there are fewer therapists trained in its application compared to more traditional therapy methods. Consequently, access can be challenging. Dr. Davis recommends searching online for clinical therapists in your area to determine whether they have been trained in VRT or VRET. Another useful resource is Virtual Reality International, which maintains a database of VRT therapists.

How Effective Is Virtual Reality Therapy?

Lucy Dunning, a licensed professional counselor in Marietta, Georgia, who incorporates VRET into her practice, notes that since the concept is still emerging, data regarding its long-term effectiveness is being developed. However, initial research indicates encouraging outcomes. “It has particularly demonstrated success for individuals with PTSD, anxiety, and chronic pain,” she remarks.

Reports indicate that virtual reality therapy in the form of VRET has a success rate ranging from 66% to 90% for individuals with PTSD when combined with cognitive behavioral therapy (CBT), based on 2022 research published in JMIR Serious Games. Additionally, it has been shown to significantly alleviate pain as an alternative to medications. A study published in the Annals of Behavioral Medicine found that burn victims, when placed in a snowy environment where they could interact with snowmen and throw snowballs, experienced a reduction in their physical pain by 35% to 50%. Scientific studies have also shown success in overcoming spider phobias and positive results for treating individuals with eating disorders.

Most current research on VRT is concentrated on VRET, and less is known about the effectiveness of therapy involving avatars in a virtual world. One study published in Frontiers in Psychiatry found that using CBT in a virtual reality context effectively treats individuals dealing with depression, who may hesitate to pursue conventional therapy. Another article in JMIR Mental Health suggests that VRT could serve as an alternative treatment method to in-person therapy for individuals experiencing social anxiety.

Are There Risks Associated with Virtual Reality Therapy?

Virtual reality therapy has several beneficial aspects, but it also comes with downsides. Although the virtual component might enhance accessibility, using it from home necessitates a computer or smart device along with a reliable Internet connection, which may not be readily available for individuals in underprivileged areas. Some people who lack technical skills might struggle to navigate VRT, and since VRT is still fairly new, finding a qualified provider can be challenging.

As with any form of therapy, the therapist plays a critical role in determining the treatment’s effectiveness, according to Dr. Davis. Especially with Virtual Reality Exposure Therapy (VRET) utilizing realistic scenarios, simulations that are too realistic could possibly cause distress for participants if they are not supported by a skilled therapist.

Another important factor is the therapist’s ability to assist the individual if an issue arises. “If the therapist is located far away, or if the person receiving counseling is interacting anonymously, this introduces significant ethical concerns and issues,” she states. “In any therapy setting, safeguarding the patient is essential. Anonymity can pose a substantial risk.” As with many new technologies, ethical considerations will likely need to be addressed as this treatment becomes more common.

The Prospects for Virtual Reality Therapy

Dr. Davis and Dunning are enthusiastic about the potential for VRT. “The future looks promising as technology evolves to become more advanced, affordable, and accessible,” Dr. Davis notes. Dunning concurs, suggesting that its usage will expand as additional VRT platforms emerge.

If you are dealing with anxiety, depression, PTSD, chronic pain, or wish to overcome a phobia and have an interest in technology and creativity, VRT might be a suitable option for you.

It is crucial to consult with a provider who is specifically trained in VRT. “Take the time to investigate which clinical practices exist in your area and whether the providers possess training in VRT,” advises Dr. Davis. Before deciding to proceed, ensure you know who is facilitating the sessions. Unlike clients attending therapy, the identity of the therapist should not remain anonymous so you can research their qualifications and confirm that they are properly trained to assist you effectively.

The expense associated with VRT varies based on the provider, the individual’s health insurance, and any required equipment (such as a headset) for at-home use. Some VRT sessions may be pricier than other types of therapy when considering the costs for equipment, as noted by the mental health non-profit Panic Anxiety Community Support. Thankfully, prices for VRT software are declining, making this therapy increasingly affordable.

Therapy through virtual reality is no longer a concept of the future; it is occurring presently. As VRT is still in its early stages, there is no existing data regarding how many people currently utilize it, yet as more clinicians gain training and research increases, accessibility will grow. “There is tremendous potential for development and learning,” states Dr. Davis.
November 18, 2024
AMD launched a new artificial-intelligence chip named MI325X

Advanced Micro Devices announced on Thursday that it intends to begin mass production of a new variant of its artificial intelligence chip named MI325X in the fourth quarter, aiming to strengthen its position in a market primarily led by Nvidia.

During an event in San Francisco, AMD CEO Lisa Su stated that the company is set to launch its next-generation MI350 series chips in the latter half of 2025. These chips feature an enhanced memory capacity and will incorporate a new base architecture that AMD claims will significantly boost performance compared to the previous MI300X and MI250X chips.

These announcements were largely anticipated due to AMD’s disclosures earlier this year. They did not inspire confidence among investors, resulting in a nearly 5% drop in AMD shares during afternoon trading. Some analysts pointed to the lack of significant new cloud-computing clients for the chips as a reason for the decline.

Shares of competitor Nvidia rose by 1.5%, whereas Intel’s shares decreased by 1.6%.

The demand for AI processors from major tech companies like Microsoft and Meta Platforms has significantly surpassed the supply available from Nvidia and AMD, enabling the semiconductor firms to sell all that they can manufacture.

This surge in demand has led to a substantial increase in chip stocks over the past two years, with AMD’s shares rising about 30% since their recent low in early August.

“There have not yet been any newly announced customers,” noted Summit Insights research analyst Kinngai Chan, who added that the stock had already increased in anticipation of “something new” before the event.

Connected to this, AMD, based in Santa Clara, California, revealed that vendors such as Super Micro Computer would start delivering its MI325X AI chip to clients in the first quarter of 2025. The design is aimed at competing with Nvidia’s Blackwell architecture.

The MI325X chip uses the same architecture as the already-released MI300X launched by AMD the prior year. The new chip features a novel type of memory that AMD states will accelerate AI processing.

AMD’s upcoming AI chips are expected to exert additional pressure on Intel, which has struggled with a consistent strategy for AI chips. Intel anticipates AI chip sales exceeding $500 million in 2024.

NEW SERVER, PC CHIPS

At the event, AMD’s Su also mentioned that the company currently has no plans to utilize contract chip manufacturers beyond Taiwan’s TSMC for advanced manufacturing processes, which are essential for creating high-speed AI chips.

“We are eager to utilize more manufacturing capacity outside of Taiwan. We are actively utilizing TSMC’s facility in Arizona,” Su remarked.
AMD also introduced several networking chips designed to enhance data transfer between chips and systems in data centers.

The company announced the launch of an updated version of its server central processing unit (CPU) design. The chip family, previously codenamed Turin, includes a variant specifically designed to ensure that the graphics processing units (GPUs) are supplied with data efficiently, which will enhance AI processing speed.

The premier chip features nearly 200 processing cores and is priced at $14,813. The entire line of processors employs the Zen 5 architecture, which provides speed enhancements of up to 37% for advanced AI data processing.

Additionally, AMD unveiled three new PC chips developed for laptops, based on the Zen 5 architecture. These new chips are optimized for running AI applications and will support Microsoft’s Copilot+ software.

In July, AMD revised its AI chip revenue forecast for the year to $4.5 billion, up from the previous estimate of $4 billion. The demand for its MI300X chips has surged due to the excitement surrounding the development and implementation of generative AI technologies.

Analysts are projecting AMD’s data center revenue for this year to reach $12.83 billion, according to LSEG estimates. Meanwhile, Wall Street expects Nvidia’s data center revenue to hit $110.36 billion. Data center revenue serves as a proxy for the AI chips required to create and run AI applications.

Rising earnings expectations from analysts have kept AMD and Nvidia’s valuations in check, despite the increase in their share prices. Both companies are trading at more than 33 times their estimated 12-month forward earnings, in contrast to the benchmark S&P 500, which stands at 22.3 times.

The Instinct MI325X, as the chip is known, is set to begin production by the end of 2024, according to Advanced Micro Devices, which announced the new product on Thursday. If developers and cloud companies view AMD’s AI chips as a close alternative to Nvidia’s offerings, it may put pressure on Nvidia’s pricing, which has maintained approximately 75% gross margins amid high demand for its GPUs over the past year.

Advanced generative AI technologies, like OpenAI’s ChatGPT, necessitate enormous data centers packed with GPUs for essential processing, prompting a demand for more firms to produce AI chips.

In recent years, Nvidia has held a dominant position in the data center GPU market, while AMD has typically ranked second. Now, AMD is striving to gain market share from its competitor in Silicon Valley or at least capture a significant portion of the market, estimating it will be valued at $500 billion by 2028.

“AI demand has significantly increased and has actually surpassed expectations. It’s evident that investment rates continue to rise everywhere,” AMD CEO Lisa Su stated during the event.

At the event, AMD did not disclose any new major cloud or internet clients for its Instinct GPUs, though the company has previously mentioned that Meta and Microsoft purchase its AI GPUs and that OpenAI utilizes them for some applications. The company also withheld pricing details for the Instinct MI325X, which is usually sold as part of a complete server system.

With the launch of the MI325X, AMD is speeding up its product release schedule to introduce new chips on an annual basis to better compete with Nvidia and capitalize on the AI chip surge. This new AI chip serves as the successor to the MI300X, which began shipping late last year. AMD indicated that its chip for 2025 will be named MI350, and its chip for 2026 will be called MI400.

The introduction of the MI325X will place it in competition with Nvidia’s forthcoming Blackwell chips, which Nvidia has announced will start shipping in substantial quantities early next year.

A successful debut for AMD’s latest data center GPU could attract investors looking for additional companies poised to benefit from the AI surge. So far in 2024, AMD’s stock has risen by only 20%, while Nvidia’s has surged over 175%. Most industry forecasts suggest that Nvidia commands more than 90% of the data center AI chip market.

AMD’s primary challenge in gaining market share lies in the fact that its competitor’s chips utilize their proprietary programming language, CUDA, which has become the standard for AI developers. This effectively locks developers into Nvidia’s ecosystem.

In response, AMD announced this week that it has been enhancing its competing software, ROCm, to enable AI developers to more easily transition their AI models to AMD’s chips, which they refer to as accelerators.

AMD has positioned its AI accelerators as being particularly effective for scenarios where AI models are generating content or making predictions, rather than when an AI model is processing large amounts of data to make improvements. This is partly attributed to the advanced memory AMD employs on its chip, which, according to them, allows it to serve Meta’s Llama AI model more efficiently than certain Nvidia chips.

“What you see is that the MI325 platform delivers up to 40% greater inference performance than the H200 on Llama 3.1,” said Su while referring to Meta’s large language AI model.

Additionally facing competition from Intel

While AI accelerators and GPUs have become the most scrutinized segment of the semiconductor sector, AMD’s primary business has revolved around central processors, or CPUs, which are fundamental to nearly every server globally.

AMD’s data center revenue in the June quarter more than doubled year-over-year to $2.8 billion, with AI chips representing only about $1 billion of that total, the company reported in July.

AMD accounts for approximately 34% of all expenditures on data center CPUs, as stated by the company. However, this is still less than Intel, which remains the dominant player in the market with its Xeon chip series. AMD aims to change this narrative with its newly introduced line of CPUs, known as EPYC 5th Gen, which was also revealed on Thursday.

These chips come in various configurations, from an economical and energy-efficient 8-core chip priced at $527 to high-end 192-core, 500-watt processors intended for supercomputers costing $14,813 each.

The new CPUs are particularly effective for supporting data feeding into AI workloads, according to AMD. Almost all GPUs require a CPU to be present in the same system to power on the computer.

“Today’s AI is largely reliant on CPU capabilities, which is evident in data analytics and various similar applications,” Su stated.

With its latest chip, AMD aims to close the performance gap with Nvidia in the AI processor sector. The company from Santa Clara also announced intentions for its forthcoming MI350 chip, designed to compete directly with Nvidia’s new Blackwell system, which is anticipated to ship in the latter half of 2025.

In a discussion with the Financial Times, AMD CEO Lisa Su articulated her goal for AMD to establish itself as the “end-to-end” leader in AI within the next ten years. “This is just the start of the AI race, not the conclusion,” she stated to the publication.

As per AMD’s website, the newly introduced MI325X accelerator comprises 153 billion transistors and is constructed on the CDNA3 GPU architecture utilizing TSMC’s 5 nm and 6 nm FinFET lithography methods. This chip features 19,456 stream processors and 1,216 matrix cores distributed across 304 compute units. With a peak engine clock of 2100 MHz, the MI325X achieves a maximum performance of 2.61 PFLOPs in peak eight-bit precision (FP8) operations. For half-precision (FP16) tasks, it reaches 1.3 PFLOPs.

A small portion of Nvidia’s AI market share

The announcement of the new chip surfaces as Nvidia’s customers prepare to implement its Blackwell chips in this quarter. Microsoft has already become the first cloud service provider to feature Nvidia’s latest GB200 chips, which integrate two B200 Blackwell chips along with a “Grace” CPU for enhanced performance.

Although AMD has positioned itself as Nvidia’s nearest rival in the off-the-shelf AI chip market, it still trails in market share, according to the Financial Times. AMD forecasts $4.5 billion in AI chip sales for 2024, a fraction compared to Nvidia’s $26.3 billion in sales of AI data center chips for the quarter ending in July. Nevertheless, AMD has already secured Microsoft and Meta as clients for its current generation of MI300 AI GPUs, with Amazon potentially following suit.

The company’s renewed emphasis on AI signifies a transition from its historically PC-centric business focusing on consumer graphics cards; however, Su remains hopeful about the rising demand for AI data center GPUs. AMD estimates that the total addressable market for AI chips will hit $400 billion by 2027.

Technological Insights on AMD’s New AI Chips

AMD’s recent AI chip, the Instinct MI325X, marks a considerable technological leap designed to contest Nvidia’s supremacy in the AI chip arena. The MI325X showcases remarkable specifications, featuring 256GB of HBM3E memory and a bandwidth of 6 TB/s, surpassing Nvidia’s H200 chip in several critical aspects. AMD claims that the MI325X offers up to 40% greater inference performance on Meta’s Llama 3.1 AI model compared to Nvidia’s H200 chip. This performance enhancement is vital as AI models grow more intricate and necessitate increased computational capability.

Along with the MI325X, AMD has unveiled the forthcoming MI350 series, which is expected to debut in the latter half of 2025. The MI350 series is projected to provide a 35-fold enhancement in inference performance over the MI300X, featuring 288GB of HBM3E memory and 8 TB/s memory bandwidth. These advancements underline AMD’s dedication to advancing the performance of AI chips and establishing itself as a strong rival to Nvidia.

Strategic Alliances and Market Dynamics

AMD’s partnerships with major technology players such as Meta, Google, Oracle, and Microsoft are essential to its strategy against Nvidia. During the Advancing AI event, AMD CEO Lisa Su highlighted these collaborations, pointing out that Meta has leveraged over 1.5 million AMD EPYC CPUs and Instinct GPUs for initiatives like its Llama large language model. These alliances not only validate AMD’s technological expertise but also create opportunities for AMD to expand its foothold in the AI market.

The AI chip sector is projected to grow to $500 billion by 2028, and AMD is eager to seize a larger piece of this lucrative market. Currently, Nvidia dominates with over 90% of the data center AI chip market; however, AMD’s assertive approach with its new AI chips and strategic collaborations suggests a strong desire to contest Nvidia’s lead. At the end of Q2 2024, AMD’s market share for EPYC server processors reached a record high of 34%, indicating potential for ongoing growth in the AI chip space.

Comparative Performance Metrics

When evaluating AMD’s Instinct MI325X alongside Nvidia’s H200 chip, several key performance metrics emerge. The MI325X yields 40% greater throughput and 30% reduced latency for a 7-billion-parameter Mixtral model, in addition to 20% less latency for a 70-billion-parameter Llama 3.1 model. Furthermore, the MI325X reportedly excels by being 10% faster than the H200 in training a 7-billion-parameter Llama 2 model. These performance metrics highlight AMD’s capability to provide competitive AI solutions that can rival those of Nvidia.

Moreover, AMD’s MI325X platform, which showcases eight GPUs, delivers 2TB of HBM3E memory and 48 TB/s of memory bandwidth, offering 80% more memory capacity and a 30% increase in memory bandwidth compared to Nvidia’s H200 HGX platform. These improvements are essential for managing extensive AI workloads and exemplify AMD’s commitment to providing high-performance solutions.

As AI technologies like OpenAI’s ChatGPT continue to create a significant need for data center processing power, AMD recognizes an opportunity to capture a considerable share of this expanding market. The AI chip sector is expected to be valued at around $500 billion by 2028, indicating immense growth potential, and AMD is positioning itself to be a key player in this arena.

Lisa Su, CEO of AMD, emphasized the rapidly increasing need for AI technology, remarking, “AI demand has outstripped expectations, and investments are growing across the board.” Although AMD did not disclose any new major cloud partnerships at the launch event, it has previously announced collaborations with Meta and Microsoft for its AI chips, and OpenAI employs AMD’s products for certain applications.

The newly introduced MI325X chip is crafted to excel in scenarios where AI models are tasked with creating content or making predictions, largely due to its sophisticated memory capabilities. AMD claims that its chip surpasses Nvidia’s H200 GPU by up to 40% when executing Meta’s Llama 3.1 AI model, representing a notable edge for specific AI tasks.

While Nvidia maintains over 90% of the data center AI chip market, AMD’s latest chip and its ROCm software ecosystem strive to facilitate AI developers’ transition from Nvidia’s proprietary CUDA programming language. This approach could assist AMD in attracting developers and companies seeking alternatives to Nvidia’s hardware.

AMD’s approach includes a quicker product release strategy, intending to introduce new chips on an annual basis. Following the MI325X, AMD plans to launch the MI350 in 2025 and the MI400 in 2026 to keep up with Nvidia’s aggressive development pace, which includes the forthcoming Blackwell chips.

In addition to its AI-targeted GPUs, AMD is reinforcing its primary CPU business. The company unveiled its fifth-generation EPYC CPUs, designed for data centers and AI tasks. These processors range from budget-friendly 8-core versions to powerful 192-core models intended for supercomputers, allowing AMD to compete with Intel’s Xeon lineup.

With AI chips representing around $1 billion out of its $2.8 billion in data center sales during the June quarter, AMD continues to challenge both Nvidia and Intel in this rapidly changing market.

The chief executive of the US semiconductor company, Lisa Su, also revealed future plans to introduce next-generation AI chips. The upcoming MI350 series chips are anticipated to be launched in the second half of next year. These chips will feature enhanced memory and an innovative architecture expected to significantly improve performance compared to the current MI300X and MI250X models.

Despite these announcements, AMD’s shares fell by nearly 5%, with some analysts linking the decline to the absence of significant new cloud-computing clients for its AI chips. Conversely, Nvidia’s stock rose by 1.5%, while Intel, another major chip player, experienced a 1.6% decrease.

The rise in demand for AI processors, driven by large tech companies such as Microsoft and Meta Platforms, has significantly surpassed supply. Both Nvidia and AMD have profited from this increase, with AMD’s stock climbing approximately 30% since early August.

AMD confirmed that vendors, like Super Micro Computer, will begin delivering the MI325X AI chip to customers in Q1 2025. The MI325X utilizes the same architecture as the MI300X chip, released last year, but incorporates new memory designed to enhance AI processing speeds.

Additionally, the company rolled out several networking chips aimed at optimizing data transfer between chips and systems within data centers. AMD also introduced a new iteration of its server CPU design. Previously codenamed Turin, the new family of chips includes a model specially designed to optimize data flow to GPUs for enhanced AI processing.

AMD also launched three new laptop PC chips based on the Zen 5 architecture, optimized for AI uses, and designed to be compatible with Microsoft’s Copilot+ software.

AMD’s AI strategy

In August, AMD announced its intention to acquire ZT Systems in a deal worth $4.9 billion, involving both cash and stock. ZT Systems is a provider of AI and general-purpose compute infrastructure for hyperscale computing companies and specializes in supplying hyperscale server solutions for cloud applications. The company has a global manufacturing presence that extends across the US, EMEA, and APAC.

AMD’s new initiatives come at a time when the semiconductor sector is facing heightened demand due to the growth of AI technologies. The rise of generative AI and advanced technologies has put pressure on supply chains as firms ramp up production of AI-focused chips. This surge in demand for AI chips raises concerns about potential shortages.

A report from Bain and Company indicates that the AI-driven spike in demand for GPUs alone could lead to a 30% or more increase in total demand for certain upstream components by 2026. Despite initiatives like the US CHIPS Act, supply limitations and geopolitical tensions may impede the industry’s capacity to satisfy demand, particularly given the complexities involved in ramping up production for advanced AI chips.

Hyperscale server solutions provider ZT Systems will be acquired by AMD in a deal valued at $4.9bn

Advanced Micro Devices (AMD) has agreed to purchase ZT Systems, a provider of artificial intelligence (AI) and general-purpose computing infrastructure tailored for hyperscale computing firms, in a cash and stock agreement valued at $4.9 billion. This amount includes a contingent payout of up to $400 million, dependent on specific post-closing milestones.

“For nearly three decades, we have transformed our business to become a top provider of essential computing and storage infrastructure for the world’s leading cloud companies,” stated ZT Systems’ CEO, Frank Zhang. “AMD shares our vision regarding the crucial role our technology and staff play in designing and constructing the computing infrastructure that powers the largest data centers globally.”

ZT Systems 101

Located in New Jersey, ZT Systems specializes in providing hyperscale server solutions for cloud computing and AI, with a worldwide manufacturing presence that extends across the US, EMEA, and APAC regions. By acquiring ZT Systems, AMD aims to enhance its AI strategy to deliver leading AI training and inference solutions through innovation in silicon, software, and systems.

Furthermore, ZT Systems’ knowledge in designing and optimizing cloud computing solutions is anticipated to assist cloud and enterprise clients in accelerating the deployment of AMD-driven AI infrastructure at scale.

“ZT brings exceptional systems design and rack-scale solutions expertise that will considerably enhance our data center AI systems and customer support capabilities,” commented AMD’s chair and CEO, Lisa Su. “This acquisition also builds upon the investments we have made to fast-track our AI hardware and software roadmaps.

“Integrating our high-performance Instinct AI accelerator, EPYC CPU, and networking product lines with ZT Systems’ top-tier data center systems expertise will empower AMD to provide comprehensive data center AI infrastructure at scale in collaboration with our ecosystem of OEM and ODM partners.”

Following the conclusion of the deal, ZT Systems will become part of the AMD Data Center Solutions Business Group. According to the semiconductor firm, Zhang will oversee the manufacturing division, while ZT Systems president Doug Huang will manage the design and customer support teams.

Additionally, AMD intends to seek out a strategic partner to take over ZT Systems’ data center infrastructure manufacturing operations based in the US. Subject to regulatory approvals and other standard conditions, the transaction is anticipated to be finalized in the first half of 2025.

AMD vs. Nvidia

AMD’s acquisition of ZT Systems signifies a strategic move to bolster its AI capabilities. This decision comes in the wake of the company’s substantial progress in AI over the course of the year, which includes its $665 million purchase of Silo AI, a Finnish AI startup.

This acquisition is part of AMD’s broader strategy to improve its position against Nvidia. The company showcased its AI initiatives at Computex 2024, where AMD presented the Instinct MI325X accelerator and announced plans for the MI350 series, projected to launch in 2025.

These advancements are integral to AMD’s plans to close the competitive gap with Nvidia in the AI semiconductor industry. Moreover, AMD has not only intensified its internal research and development (R&D) activities but has also put over $1 billion into expanding its AI ecosystem and enhancing its AI software capabilities over the past year.

AMD’s CEO, Lisa Su, informed Wall Street analysts that interconnected server racks utilizing tens of thousands of GPUs for model training and inferencing are expected to become increasingly intricate over time. Consequently, customers will require a chip vendor capable of assisting them in designing systems and expediting production.

Presently, organizations usually take several quarters from the initial sampling of GPUs to deploying them within servers that handle production workloads, Su noted.

“The ZT team will assist AMD in scaling up rapidly,” Su mentioned during the conference call with analysts. “We can effectively conduct a substantial amount of development concurrently.”

ZT will aid AMD’s largest clients in developing their AI infrastructure. Simultaneously, the chip manufacturer will fine-tune its GPUs and CPUs for these systems, according to Su. Nevertheless, ZT will continue to create systems for entities looking to use silicon from rival companies.

“This initiative will not limit customer choice,” Su stated. “Some hyperscalers will seek distinct system design optimizations, and we will have the team available for that.”

AMD is significantly smaller in the AI accelerator market compared to Nvidia. Nvidia reported $22.6 billion in data center revenue for the quarter that concluded in April, with a considerable share derived from AI systems. AMD anticipates $4.5 billion in sales this year from its AI data center GPUs.

ZT also designs and produces non-AI CPU-based systems, suggesting that the acquisition could enhance AMD’s competitiveness against Intel in large organizations’ data centers, said Jack Gold, an analyst at J.Gold Associates. AMD could leverage ZT to promote its EPYC CPU in competition with Intel’s Xeon chip.

“With ZT providing non-AI solutions as well, this represents a direct challenge to Intel from AMD,” Gold commented on LinkedIn.

Analysts predict that the demand for AI GPUs will surpass that of CPUs in extensive data centers. AMD is rapidly launching AI accelerators to expand its market share, which Su believes will grow from $45 billion last year to $400 billion by 2027.

In December, AMD introduced the MI300 Series, marking its inaugural Instinct AI accelerator for hyperscale data centers. In 2026, the company intends to release the MI400 series aimed at large-scale AI training and inferencing. For programming GPUs that run large language models, AMD provides its ROCm open software stack consisting of tools, libraries, and APIs.

AMD plans to divest ZT’s hardware manufacturing division after finalizing the acquisition, Su indicated. ZT’s revenue was approximately $10 billion over the past year, primarily from its manufacturing division.

AMD expects to keep around a thousand of the privately held company’s 2,500 employees, anticipating operating expenses of $150 million. The chipmaker expects ZT to start contributing to its gross revenues in 2026.

Post-acquisition, ZT will integrate into AMD’s data center business group. ZT CEO Frank Zhang will take charge of AMD’s manufacturing operations, while ZT President Doug Huang will lead AMD’s system design teams.

The ZT acquisition followed closely after AMD completed the $665 million purchase of Silo AI, a European lab focusing on AI services for autonomous vehicles, manufacturing, and smart cities.

This ZT acquisition is among AMD’s most significant. In 2022, AMD purchased Xilinx, known for programmable integrated circuits that customize microprocessors, for $50 billion. That same year, AMD also acquired Pensando for $1.9 billion, which developed programmable data processing units to alleviate CPU workloads on servers.

Frank Zhang, who founded and leads ZT Systems as the CEO, will keep managing the manufacturing division and fulfill the commitments to current clients after AMD finalizes its acquisition of the company, expected to be completed early next year. In the interim, Zhang will seek a buyer for the manufacturing operations, which employs about 1,500 people, since AMD is not interested in competing with its customers by engaging in server manufacturing and sales. This stands in contrast to another well-known GPU system manufacturer we recognize.

Additionally, AMD has already experienced this with the microserver pioneer SeaMicro, which it acquired in March 2012 for $334 million under the leadership of CEO Rory Read (remember him?), just as Lisa Su transitioned from IBM Microelectronics to lead its global business units. They eventually shut it down in April 2015 as AMD reset its server business after Su took over as president and CEO.

“Clearly, we have already started discussions with all our OEM and ODM partners,” says Forest Norrod, general manager of AMD’s datacenter business and formerly in charge of custom server business at Dell, in an interview with The Next Platform. “A reassuring factor is that all of these discussions have been very positive. People quickly understand the rationale behind our decision, and they recognize and appreciate that we have no plans to compete with them. We’re not going to do that, it’s not going to happen. I fully understand both businesses and there’s no confusion on my part.”

AMD aims to enhance its systems architecture and engineering capabilities. Currently, AMD has approximately 500 system engineers, according to Norrod, whereas ZT Systems has 1,100 individuals performing this work. Since AMD designs systems according to multiple standards rather than just one, it requires a larger workforce to assist in the design and development of future GPU-accelerated systems, which will pose challenges; however, they do not plan to engage in production manufacturing.

It remains uncertain what AMD will acquire with the divestiture of ZT Systems’ manufacturing business, but acquiring 1,100 experienced system engineers would be prohibitively costly and might not be feasible through any other means than acquiring a specialized high-performance system manufacturer such as ZT Systems.

This option is more economical than buying Supermicro, and likely offers a similar number of system engineers.

Here’s the situation as Norrod explains it, and we provide the full quote to illustrate AMD’s reasoning for investing $4.9 billion in ZT Systems, which amounts to $4.45 million for each system engineer. (Some costs will be offset by the divestiture of the manufacturing side, of course.) Here is how Norrod articulated it:

“We have been looking ahead at the roadmap and grasping the challenges of designing competitive systems that excel in performance and efficiency. With the rise of AI systems, it’s becoming increasingly obvious to everyone in the sector that this will lead to significant challenges in designing systems capable of operating at these power levels and signaling rates, given the complexity involved. Maintaining and managing these systems will be quite challenging.”

“There are numerous issues that need addressing, and the requirements to meet these challenges trace back to the very early stages of the silicon development process. We are acquainted with some of these challenges since they are typical in supercomputer design. However, when examining the developments within AI systems, the complexity is increasing rapidly, making it essential for us to have a sufficient number of world-class system design engineers involved right from the silicon definition stage. Thus, it became apparent that we needed to significantly improve our capabilities here.”

“Furthermore, as we enhance our capabilities, we want to remain true to AMD’s legacy of fostering open ecosystems and offering customer choices, rather than constricting them within proprietary confines. Consequently, this necessitates an even larger number of engineers. If you wish to create a single proprietary system for universal use, you require a certain staffing level. However, to develop open ecosystems that accommodate choice and variation entails greater complexity and requires additional system engineers to ensure timely market delivery and uphold high quality.”

This is largely about accelerating time to market and enhancing the system design and engineering capabilities. AMD has effectively developed impressive CPUs and now GPUs, but it must create a comprehensive networking stack and system boards that integrate well with rackscale and cluster-wide system architectures, which should be thoroughly tested and validated at scale. This is the reason Nvidia established the DGX series, and AMD acknowledges that this is necessary, yet it will not manufacture systems for customers nor take on the role of a prime contractor for HPC or AI clusters. This is in contrast to Intel’s attempts, which did not succeed very well.

AMD’s acquisition of ZT Systems involves purchasing ZT’s rack-scale systems design and manufacturing assets for $4.9 billion, with 75% paid in cash and 25% in stock. This transaction builds on the $1 billion AMD has already invested in ZT over the previous year.

The acquisition will primarily focus on the design of Instinct AI GPUs, EPYC CPUs, and networking solutions from AMD and its partners. AMD plans to divest ZT’s manufacturing assets, retaining the systems design capabilities.

Frank Zhang, the CEO and founder of ZT, will oversee the manufacturing division that will be sold, while ZT President Doug Huang will manage design and customer enablement, reporting to Forrest Norrod, who leads AMD’s Data Center Solutions Business Group.

The AMD board has approved the deal, which is anticipated to finalize in the first half of 2025, pending regulatory approvals. It is expected that the acquisition will be beneficial for AMD on a non-GAAP basis by the end of 2025.

ZT Systems is engaged in the design, integration, manufacturing, and deployment of rack-level, hyperscale AI systems. It is rumored to generate $10 billion in annual revenue, mainly from its largest clients, AWS and Azure.

The company employs approximately 1,000 personnel in design and customer roles and another 1,000 in manufacturing. Founded in 1994 and based in Secaucus, New Jersey, ZT has evolved from producing desktop PCs and pedestal servers in its early days to focusing on data center servers since 2004, then transitioning to rack-scale design and integration in 2010, followed by a commitment to hyperscale solutions in 2013, and in 2024, it will ship “hundreds of thousands of servers annually.”

AMD’s acquisition of ZT positions the company for significant growth in the datacenter AI market. The increase in sales of AMD’s Instinct GPUs has been substantial, showing growth from $0 in the first half of 2023 to a projected $4.5 billion by 2024, driven by considerable investments in hardware and software. However, in comparison to AMD’s own AI accelerators and GPUs market forecast of $400 billion by 2027, the company requires catalysts to facilitate its rapid growth and capture what I refer to as its “unfair” share.

Although there have been improvements, AMD faces two primary competitive hurdles in AI infrastructure: its software limitations and the scale and maturity of its systems. While AMD has effectively addressed these issues for non-AI EPYC servers and PCs, its solutions for AI racks require enhancement. AMD could develop these capabilities in-house, but the time required for such an endeavor is considerable.

The company has previously completed three minor software acquisitions (Silo AI, Nod.ai, and Mipsology) to enhance its mid- and high-level software functionalities and support customer customization of LLMs. Furthermore, AMD has made significant progress with ROCm AI optimizations and compatibility with PyTorch and Hugging Face for both Instinct and EPYC. I anticipate that AMD will pursue additional software acquisitions in the future.

While AMD could not foresee a $4.5 billion annual projection for Instinct without some systems capabilities, what it currently possesses is insufficient to carve out its fair share of the anticipated $400 billion market. The AI infrastructure landscape extends beyond merely being a chip-focused arena; it has transitioned to encompass a more integrated system and software approach. “Chip” manufacturers are now expected to supply complete rack solutions and software stacks to achieve continuous improvements in performance, efficiency, quality, and time-to-market. The acquisition of ZT is strategically aimed at enhancing AMD’s capabilities “above the chip” and “below software” for AI server solutions.

I believe this acquisition, if executed with Lisa Su’s usual precision, will serve as the catalyst needed for AMD to drive remarkable revenue growth for both Instinct and EPYC at the head node, particularly with hyperscalers, tier-two CSPs, and some of the largest on-premises facilities for governmental, financial, energy, and pharmaceutical sectors.

I am optimistic about the cultural compatibility between the two entities. During a discussion with her regarding the deal, Su highlighted the long-standing partnership between AMD and ZT. “Our team has collaborated with them for many years,” she stated. “They contributed to some of our initial EPYC designs and MI 250 designs, and have been actively involved in MI 300 designs. This has allowed us to become very familiar with them.”

This synergy extends to customer relationships as well. Su mentioned Frank Zhang’s focus on the datacenter and cloud markets for over 15 years. Consequently, instead of spreading too thin, ZT has strategically focused on a select few crucial clients. While Su could not disclose any customer names due to ZT being a private entity, she emphasized that “Every one of their customers is our customer.” Hence, even though integrating engineering teams from different companies typically presents challenges, it is favorable that all parties will continue to serve the same clientele.

Lastly, I appreciate the decision to eventually divest ZT’s manufacturing, sales, and support functions, as these areas would dilute AMD’s focus and profitability. For context, Supermicro operates with net margins in the mid-single digits, while AMD maintains margins around 25%. In connection to this, Su mentioned that AMD would avoid entering the systems business as Nvidia has done with DGX. I have mixed feelings about this since DGX provides Nvidia with significant revenue and profit margins, creating a solid revenue stream.

Avoiding Involvement in the Systems Manufacturing Sector

At the same time, I appreciate the choice to eventually eliminate ZT’s manufacturing, sales, and support functions, as these areas would be significantly dilutive. For context, Supermicro’s net profit margins fall within the mid-single digits, while AMD’s are around 25%. In relation to this, Su mentioned that AMD would not venture into the systems market the way Nvidia has with DGX. I have mixed feelings about this, as DGX generates considerable revenue and profit for Nvidia and provides a platform to promote an all-inclusive solution. Undoubtedly, hyperscalers and top-tier OEMs would prefer AMD not to enter the systems space, but AMD needs a compensation model for avoiding competition with its clients. So far, it appears that Nvidia isn’t adversely affected by this situation.

Su believes that clients appreciate having options and customized solutions instead of a model that imposes a specific configuration of CPU, GPU, and networking within a set form factor for data centers. According to Su, this new agreement will change that perception. “We’re going to say, ‘I would welcome you to use my CPU and GPU along with our open networking standard, but actually, I will customize the system for you. Please let me know what your ideal data center would look like.’”

There’s another competitive aspect to consider. ZT Systems designs, manufactures, and implements Nvidia systems—allegedly for AWS and Azure as well. In that context, the implications of the ZT acquisition for AMD’s leading datacenter AI chip rivals, Nvidia and Intel, are somewhat unclear in terms of design. After the deal concludes, I would anticipate that all design work for Nvidia and Intel will cease. AMD claims that production for the competing systems will carry on, which seems logical if the manufacturing segment is indeed separated and sold off.

Although some may justifiably critique AMD in certain areas, pinpoint execution has become a defining characteristic of Su’s leadership. Meticulous execution is precisely what is required to turn this investment into a success for the company by boosting revenue and gaining market share. When compared to AMD’s acquisition of Xilinx a few years back, this one appears straightforward. This acquisition further solidifies the advantage that AMD and Nvidia have accumulated over their competitors in the realm of AI chips. I am confident that this purchase will be beneficial for the company and allow it to capture a larger portion of the projected $400 billion datacenter AI market by 2027.

November 17, 2024
A high-tech Electric hydrofoil ferry service began operating in Stockholm

A new electric ferry service using advanced technology started operations in Stockholm on Tuesday, providing commuters with an eco-friendly option to navigate the waters of the Swedish capital, situated across 14 islands.

In what Stockholm declared to be a world first, 25 commuters in the Ekero suburb boarded the Nova, a hydrofoil ferry that runs on electric motors. The ferry glided about 1 meter (3 feet) above the surface and traveled 15 kilometers (9 miles) to reach Stockholm’s City Hall in just half an hour. In contrast, the regular diesel-powered ferry service takes 45 minutes during the morning commute, without any stops.

“We aim to lead the way in the green transition on the water,” stated Gustav Hemming, the city councilor responsible for climate and infrastructure. The goal of the nine-month pilot project was to encourage more people to leave their cars at home and opt for a public transportation card instead.

Gustav Hasselskog, the CEO of electric boat manufacturer Candela, referred to it as “a significant change for urban transportation and a revitalization of our waterways.”

He noted that the Nova is the inaugural vessel of Candela’s new P-12 model to be put into service. Its computer-controlled hydrofoil wings elevate the hull above the water, resulting in an 80% reduction in energy consumption compared to traditional vessels by minimizing water resistance.

“Traditional ships haven’t advanced significantly in the last century and rank among the least energy-efficient modes of transport, second only to a battle tank,” Hasselskog remarked in a statement.

As of Tuesday, the Nova has officially joined the fleet of ferries managed by Stockholm’s public transport authority, SL.

The ferry is built to accommodate 25 passengers, including a wheelchair space. There are speed restrictions on certain sections of the route, but there are no limits on open water. The hydrofoil boat maintains a cruising speed of roughly 25 knots (46 kph or 29 mph) and can achieve a top speed of 30 knots (56 kph or 35 mph) — significantly faster than other electric passenger ferries. It accomplishes this with carbon fiber hydrofoil wings that elevate the vessel, reducing drag.

Additionally, the vessel is exempt from Stockholm’s 12-knot speed limit as it produces no wake — the waves created by a boat moving through the water that increase with speed and can potentially overwhelm other vessels or erode the shore.

Candela claims that its technology lowers the energy consumption per passenger-kilometer by 95% when compared to diesel ferries operating in the scenic Stockholm archipelago.

The ferry can operate in waves up to 2 meters (6.5 feet).

Candela envisions that, alongside Stockholm, cities like San Francisco, New York, and Venice will spearhead the movement towards electrifying waterborne public transportation.

Stockholm currently has about 70 public transport vessels that operate on fossil fuels. In 2022, there were approximately 6.2 million public transport boat trips in the Stockholm region, and while boat traffic constitutes a small portion of the overall public transit system, it is the fastest-growing mode of transport following the COVID-19 pandemic.

Numerous cities globally regard clean and effective public transportation as an essential method to reduce carbon emissions. For urban areas with waterways, a cutting-edge ferry in Sweden may soon establish a new benchmark.

Traveling through Stockholm’s archipelago, the new P-12 vessel from electric boat manufacturer Candela glides silently over the water at a height of about one meter (3 feet). Its creators aspire for this ferry, revealed this week, to initiate a new age of water-based public transportation.

“This represents a significant advancement,” stated Erik Eklund, who oversees Candela’s commercial vessel division. “The energy efficiency gained from flying on the foils provides us with the velocity and range required to operate on batteries.”

The vessel is engineered to transport 30 passengers, reaching a top speed of 30 knots (56 kph or 35 mph) — notably quicker than other electric ferries. It achieves this speed using carbon fiber hydrofoil wings that elevate the craft above the surface, minimizing drag.

According to Candela, this technology decreases the energy consumption per passenger-kilometer by 95% in comparison to the diesel ferries currently moving passengers across the beautiful Stockholm archipelago, which consists of countless islands and skerries extending into the Baltic Sea.

Additionally, the ferry is exempt from the 12-knot speed limit in Stockholm because it creates no wake — the waves caused by a boat moving through the water that intensify with speed and could potentially inundate other vessels or erode the shoreline.

The P-12 is still undergoing testing but is scheduled to begin operating in July on the route between the Stockholm suburb of Ekero and the city center as part of a nine-month trial project. This ferry will reduce travel time from Ekero via standard public transport from 55 minutes down to 25 minutes.

The company aims to utilize insights gained from the launch of its smaller electric hydrofoil recreational boat. Engineers onboard are refining the hydrofoils, which are adjusted by a computer 100 times every second to adapt to sea conditions and counter any wave impacts. The vessel can function in waves of up to two meters (6.5 feet).

Candela envisions that, alongside Stockholm, other cities like San Francisco, New York, and Venice will spearhead the transition to electrified marine public transport.

Gustav Hemming, Vice President of the Regional Executive Board in Stockholm, expressed support for this initiative.

“The goal is for the Stockholm region to enhance public transportation by water, as we believe it is a key factor in making public transit more appealing,” he remarked.

In 2022, there were approximately 6.2 million public transport boat trips in the Stockholm region; while boat traffic still represents a minor portion of the overall public transit system, it has become the fastest-growing form of public transportation post-COVID-19 pandemic.

“Our roads are often congested, and constructing new ones is quite costly and not very eco-friendly,” Hemming noted while gazing over Stockholm’s open waters on a chilly autumn day. “However, we have our established infrastructure here. There is no congestion on the water.”

Using hydrofoils to elevate a vessel above the water to reduce drag is not a novel concept. Ship designers have been exploring this technology for over a century, but high costs and maintenance challenges had hindered its widespread use. Nevertheless, the advent of lightweight carbon fiber materials has revived this technology in elite sailing, and with the efficiency of electric motors and the rising costs of traditional fuels, it is experiencing a resurgence in the public transport arena.

“We understand that marine vessels tend to be energy-intensive, and the restricted energy density of current batteries constrains the electrification of marine fleets,” stated Arash Eslamdoost, an associate professor of applied hydrodynamics at Chalmers University of Technology in Gothenburg. “This is where foiling presents a transformative solution to optimize the limited onboard electric power.”

Worldwide, several hydrofoil electric passenger ferries are either being designed or actively developed. In the U.K., Artemis Technologies has announced its intention to create a fully electric hydrofoil ferry to operate in Northern Ireland between Belfast and nearby Bangor, potentially launching as soon as next year.

Robin Cook from the Swedish Transport Agency noted that the maritime industry is poised for transformation, particularly in short-distance routes. However, he emphasized the need for public infrastructure to keep pace with these advancements and encourage them through incentives.

“A crucial element of electrification is when vessels connect to ports through onshore power supplies,” he commented. “In this regard, harbors play a vital role in ensuring that infrastructure is developed for these connections.”

The revolutionary ferry employs hydrofoil wings controlled by computers that elevate the hull above the water, leading to an 80% reduction in energy usage compared to traditional boats. “This represents a significant transformation for urban transportation and a rejuvenation of our waterways,” stated Gustav Hasselskog, CEO and founder of Candela.

Presently, Stockholm’s roughly 70 public transport vessels consume more fossil fuels than the city’s buses and trains combined. The new ferry tackles several challenges by running on 100% renewable electricity and producing minimal wake, enabling greater speeds within city boundaries.

Nova travels at 25 knots, making it the fastest electric ferry currently in service and beating the speeds of earlier diesel-powered boats. For Ekerö, the island suburb of Stockholm that is experiencing the most growth, this translates to a reduction in travel times from one hour to just 30 minutes. The ferry’s advanced technology features electric C-POD motors with no mechanical transmission, allowing for nearly silent operation even at top speeds. It also needs very little infrastructure, charging at a standard car fast charger during normal breaks.

The pilot initiative, executed by Candela, Trafikverket, and Region Stockholm, will run through the fall of 2024 and restart in spring 2025. Its objective is to showcase how hydrofoil technology can enhance maritime transit efficiency and environmental sustainability.

“In numerous cities, the quickest route is over water, which is humanity’s oldest infrastructure,” said Hasselskog. “Today, our waterways are not fully utilized because of high costs, concerns over wake, and the emissions associated with traditional vessels.”

The initiative has already attracted global interest, with Candela securing orders from Saudi Arabia, New Zealand, and Berlin prior to the official launch.

“This signifies a major change in urban transportation and a revival of our waterways,” commented Gustav Hasselskog, the founder and CEO of Candela, who united specialists in hydrodynamics, software, advanced computer simulations, and mechatronics in 2014 to revolutionize electric boating.

Last year, the company raised $20 million to scale up production of their ferries and water shuttles, followed by an additional €24.5 million in March 2024, which included funding from Groupe Beneteau, the largest boat manufacturer globally.

Half the commuting duration, with zero emissions Dateline: Tuesday, October 29, 2024, 07:15 CET. “Nova,” the inaugural vessel of Candela’s P-12 model, left its dock in Ekerö, the fastest-growing island suburb of Stockholm, and completed a 15 km journey to Stockholm’s City Hall in just over 30 minutes. That’s roughly half the time needed for the trip on a diesel-powered ferry. Half the time with zero carbon and toxic fume emissions.

This is achievable due to Candela’s technology, which enables the 12-passenger, 12-meter (≈ 40 ft) boat to fly a meter above the water’s surface, supported by computer-controlled hydrofoil wings that nearly eliminate water friction and decrease energy consumption by 80% compared to standard hulls.

Stockholm is located in an archipelago of 30,000 islands and has a substantial fleet of public transport vessels—about 70 in total—that utilize more fossil fuels than the city’s buses and trains combined, despite making up only a small percentage of overall transit usage. Water transport is expensive and time-consuming, as the vessels produce significant wakes, which restrict their speed in the city center.

Powered by 100% renewable electricity Since the P-12 glides above the water rather than forcing its way through like conventional vessels, it creates negligible wakes. Consequently, the ferry is permitted to operate at high speeds within city limits, where other vessels face wake restrictions.

Did we mention its quietness? Nova operates solely on renewable electricity and produces minimal noise even at high speeds, thanks to Candela-engineered electric C-POD motors with no mechanical transmission. The motors represent a notable advancement in electric boating, utilizing two motors and contra-rotating propellers housed in a torpedo-shaped casing to enhance efficiency further.

Electric hydrofoil ferries on the rise Several companies have developed hydrofoiling electric ferries and other commercial passenger ships, including MobyFly in Switzerland, Vessev in New Zealand, and Artemis Technologies in the UK. In the past year, Candela has received orders for the P-12 from Berlin, Saudi Arabia’s NEOM project, and an environmentally sensitive lake in New Zealand, with more clients expected to be announced.

Tuesday’s ‘flight’ represents the world’s first operational service for electric foiling ferries. For Stockholm—perhaps for the entire world—this indicates a future in which water transport in cities is sustainable, cost-effective, and faster than commuting by car.

Candela’s Hasselskog states, “In many cities, the most direct path is over water, which is humanity’s oldest infrastructure. Today, our waterways are underutilized due to high costs, wake-related concerns, and emissions from conventional vessels. If we can harness this potential, we can enhance the appeal of our cities.”

The Candela P-12 does not require expensive docking infrastructure. It can be charged at a standard car fast charger located at Stockholm City Hall. Its extensive range enables it to keep pace with traditional diesel ferries, allowing for recharging during the usual lunch break.

Nova will operate until the waters freeze in fall 2024, with services resuming in spring and continuing through August 2025. This route serves as a pilot initiative led by Candela, Trafikverket, and Region Stockholm (SL) to investigate how hydrofoil technology can facilitate quicker, more cost-effective, and emissions-free maritime travel, thereby creating new transit patterns in Stockholm.

Candela’s advancements in hydrofoiling are also applicable to leisure boats. Their first electric hydrofoiling vessel debuted in 2016, and the Candela C-8, which launched in 2022, has become one of Europe’s top-selling boats—regardless of propulsion—over the last few years.

A few weeks ago, the C-8 achieved a milestone by becoming the first all-electric boat to successfully cross the Baltic Sea.

From the outset, Candela’s primary goal has been “to accelerate the transition to fossil fuel-free lakes and oceans. By developing electric hydrofoil vessels that outperform fossil fuel alternatives, we’re leading the charge for zero-emission marine transportation.”

Discussing this significant advancement in waterborne public transport, Hasselskog states, “For the first time, we have a vessel that makes water transport more rapid, environmentally friendly, and cost-effective than land transportation. It heralds a new era for global waterways, and it’s thrilling that Stockholm is pioneering this change.”

Merging long-distance capability with high speed, the new Candela P-12 electric hydrofoiling water taxi takes the innovative technology from the striking Candela 7 leisure boat and adapts it for commercial operators in a vessel designed for 12 passengers.

The technology and software behind Candela’s hydrofoiling boats began development in 2014, created by a skilled team with expertise not only in boat design and hydrodynamics but also in fields like avionics, image and signal processing, dynamic modeling, control theory, and machine learning. They developed a unique flight controller/software/sensor system that collects information from the hydrofoil wings as the vessel moves and makes adjustments over a hundred times each second.

Hydrofoiling uses one-fifth the energy of traditional hulls

This leads to an exceptionally smooth experience, resembling flying more than boating, and the secret to achieving rapid movement through water over sustained periods is to elevate the vessel above it! The energy efficiency gained by minimizing hull drag through hydrofoil wings provides an unparalleled combination of speed and range.

This energy efficiency is what makes a hydrofoiling water taxi attractive to commercial operators. Candela has compared its new 8.5-meter P-12 against conventional fossil fuel water taxis across various criteria, revealing that the foiler requires only 44kW of power compared to 258kW for a similar non-hydrofoil hull. This results in significantly lower fuel expenses, especially when utilizing inexpensive electricity.

It’s well-known that the primary barrier for consumers considering an electric boat is the hefty initial cost of the battery. For commercial operators, however, the battery cost is less concerning. The more frequently the vessel is utilized, the quicker the return on that initial investment due to the lower ongoing fuel costs.

Candela estimates that operating a typical fossil fuel water taxi costs around €5 per hour (US$ 6.10), while the operating cost for a Candela P-12 is just one-fifth of that: €1 (US$ 1.22). If a boat operates with 12 paying customers for six hours a day, it quickly becomes evident how that cost difference can cover the battery expenses and subsequently contribute to profits.

Candela’s hydrofoiling water taxi is the company’s third model. Although Candela’s electric hydrofoiling provides an exciting and smooth experience (the P-12 can reach a maximum speed of 30 knots / 55 km/h), the main motivation behind developing the technology was different. The company’s primary goal “is to accelerate the shift to fossil fuel-free lakes and oceans. By developing electric hydrofoil boats that outperform fossil fuel alternatives, we’re making strides toward zero-emission marine transportation.”

Candela invested over 10,000 hours in simulations, design iterations, and sea tests to perfect its system, but a significant benefit is that they don’t need to redesign the hydrofoil for various boat sizes.

Launched in 2019, the Candela 7 recreational speedboat has garnered significant demand across Europe and North America. They announced in March their collaboration with Stockholm city to create a hydrofoiling ferry with a capacity for 30 passengers (the P-30), and now they have unveiled the hydrofoiling water taxi at the Salone Nautica in Venice, a location known for numerous commercial passenger vessels.

Typically, a personal speedboat, a 12-passenger taxi, and a 30-passenger ferry would necessitate completely different hull shapes and designs to optimize their movement through and performance on the water. However, this is not the case with hydrofoiling.

The P-12 hydrofoiling water taxi and the P-30 hydrofoiling ferry can be seen side by side. While there are variations in the size of motors and batteries needed for different Candela vessels, the core components—Candela wings, sensors, and stabilizing software—are fundamentally the same. The different hull dimensions and seating options are then customized and integrated. The P-12 shares various parts with its larger counterpart, the P-30, including the ‘climate shell’ passenger section offering 360º views.

Water taxis and small sightseeing vessels operate globally, transporting diverse groups of people to various destinations for a multitude of purposes, and an electric hydrofoiling model could enhance the experience in nearly all instances.

Where speed is a priority, the Candela P-12 delivers it—while producing less noise, minimal wake, zero emissions, and a smoother ride for passengers. During excursions in environmentally sensitive regions, the electric boat enables passengers to closely observe marine wildlife with minimal disruption to their habitat.

In the realm of electric vehicles, larger commercial units like buses, delivery trucks, and work vehicles are accelerating the transition from fossil fuels due to their significant economic advantages. The Candela P-12 provides similar advantages for commercial boats, making the shift from internal combustion engines to electric a straightforward choice, with no compromise between cost savings and environmental benefits.

Pioneering manufacturer Candela has successfully completed the largest funding round in its history, raising 25 million euros (approximately $27.13 million) to boost the production of its groundbreaking P-12 ferry.

Groupe Beneteau, the top boat manufacturer globally, is a key partner in this funding round. Their brand portfolio includes Jeanneau, Prestige, Lagoon, Wellcraft, Scarab, along with four brands already in the electric boating sector: Delphia, Four Winns, Excess Catamarans, and Beneteau itself.

Additional investors in this funding round include longstanding supporters EQT Ventures, the venture arm of Swedish firm EQT Partners; Kan Dela AB; and Ocean Zero LLC, which also funds seven other companies focused on reducing emissions in the marine industry, including ZEN Yachts and Flux Marine. This fresh capital injection brings Candela’s total funding since its inception to over €70 million (around $76 million).

Founded in 2014, Candela originated in Sweden when Gustav Hasselskog envisioned creating an electric boat that could match the range and speed of gasoline-powered vessels. He assembled a cutting-edge team with expertise across hydrodynamics, flight control electronics, structural composite engineering, and the software utilized for dynamic modeling.

They developed the first Candela hydrofoiling speedboat along with an onboard flight controller that collects real-time data from sensors located around the boat and adjusts the foils over 100 times each second to counteract wave and water movements.

While experiencing hydrofoiling in a recreational vessel like the Candela 8 is certainly extraordinary and thrilling, the more significant environmental impact will arise from Candela’s fleets of water taxis and larger ferries. In Europe, ferries contribute to 10% of CO2 emissions from all shipping vessels.

Candela’s boats are designed to consume 80% less energy compared to others due to their nearly negligible water resistance and friction. This technology reduces lifetime emissions by 97.5% when contrasted with diesel-powered vessels, all while allowing operators to cut their costs by half. Since it produces minimal wake, the P-12 has received exemptions from speed regulations. In Stockholm, it will reduce travel times to half of what they are with traditional road transport and older diesel-burning boats.

The electric vessel market is projected to reach a value of 14.2 Billion USD by 2030, as reported by Fortune Business Insights, fueled by strong governmental incentives aimed at decarbonizing shipping.

Bruno Thivoyon, CEO of Groupe Beneteau, stated, “Our investment aligns perfectly with Groupe Beneteau‘s sustainability objectives, enhancing innovative solutions for more eco-friendly boating and exceptional experiences. Candela’s technology, which allows for significantly more efficient electric vessels, will revolutionize waterborne transport in its next sustainable phase.”

Groupe Beneteau stands as the largest boat manufacturer globally, boasting a turnover of €1.46 billion, with 15 factories, 9 distinct brands, and over 8,000 yachts produced each year.

In 2021, they announced their own ventures into electric recreational boating during the Cannes Yachting Festival, introducing Delphia as their dedicated electric brand. The inaugural Delphia 11 cruising boat was launched in Europe in 2022, and another brand, Four Winns, unveiled the electric H2E sportboat in 2023.

“Charting the path toward a brighter future,” said Candela founder Hasselskog. “We are thrilled to have Groupe Beneteau on our team. As the leading global boat manufacturer, their endorsement is a strong confirmation of our technology’s capability to transform waterborne transportation. We are eagerly looking forward to the opportunities that lie ahead.”

He also expressed this sentiment to those who have supported Candela from the outset:

Today, we revealed our latest funding round of €25 million – the most substantial in Candela’s history.

We wouldn’t have achieved this milestone without you – our remarkable customers, investors, and partners. Our talented team has dedicated countless hours to develop our innovative electric hydrofoil vessels from a mere concept to prototypes, and then to best-sellers in the electric leisure sector, now expanding into the commercial passenger transport arena with the Candela P-12.

This funding round also signifies a new pinnacle as we welcome Groupe Beneteau as investors. Having the largest boat manufacturer in the world on board serves as a significant endorsement of the transformative potential our technology holds for global waterborne transportation.

Waking up to this news feels surreal; I often need to pinch myself to believe that my nearly 10-year-old aspiration of creating the first practical, long-range electric boat has materialized. We’re now positioned to guide the future, charting a course toward a more sustainable tomorrow.

A sincere THANK YOU goes out to everyone who believed in this vision and joined us on this incredible journey. Stay tuned for what lies ahead—we’re only just beginning!

Denison is the US company uniquely positioned to take on the Candela 7, as both firms were established by visionaries motivated by the thrill of innovating and seizing overlooked opportunities.

Frank Denison began his journey by mending and reselling boats purchased on speculation, and is recognized for being the first to install diesel engines in a yacht during the mid-1930s. His grandson Bob, who currently leads Denison and transitioned the company to a yacht brokerage in 2009, describes Frank as “an incredible boat builder” who also introduced the first turbine-powered yacht in 1973. Frank’s wife, ‘Kit’ Denison, created the first ‘country kitchen’ galley for yachts in the 1980s.

Gustav Hasselskog, a pioneer in electric boating at Candela, expresses his ambition to “revolutionize the industry – eliminating reliance on fossil fuels in boating.” After exploring the various possibilities within electric boating and examining the engineering and physics of the challenge, he concluded that the most effective way to demonstrate the advantages of electric boats was to design one that could fly. He assembled a team of specialists in flight control electronics, software algorithms, hydrodynamics, and structural composite engineering to achieve this.

While there are many excellent electric boats that plane on the water in the conventional manner and offer outstanding environmental performance using electric motors instead of gasoline, Haaselskog chose a different strategy by pursuing hydrofoiling due to a significant reason: water has a higher density than air. Regardless of how light a hull is made or how aerodynamic its shape is optimized, substantial energy is required to overcome the drag and friction of a boat hull moving through water. Candela’s research demonstrated that “a 7.5 meter (24 ft) planing boat consumes 12-18 times more fuel than a family car.”

Hydrofoiling is not a novel concept; in fact, the earliest evidence of hydrofoils on vessels appears in a British patent awarded to Emmanuel Denis Farcot in 1869, and military boats have utilized them since World War II. The principle remains unchanged: a boat reaches a certain speed at which it lifts out of the water, riding on a slightly submerged T or V-shaped wing, resulting in minimal drag and resistance.

Candela has elevated the concept to an entirely new level suited for the 21st century, made possible by their diverse team of experts. Weight is clearly a crucial element in hydrofoiling; thus, the hull, deck, and all deck components of a Candela 7 are constructed from carbon fiber, showcasing the structural composite engineering aspect.

The expertise in flight control electronics and software management has been integrated into the hydrofoil design. Unlike traditional hydrofoils that are static, the Candela adjusts dynamically in electric flying boat mode, as seven sensors continuously gather data on the boat’s position, velocity, and acceleration across the x, y, and z axes, along with its rotational movement. This information is relayed to the flight control software, which constantly adjusts the wings to maintain optimal height, roll, and pitch.

You may wonder what occurs with the foils in shallow water. The foils are designed to retract completely into the hull, and the motor can be tilted upwards, resulting in a draft of merely 0.4 meters / 1 foot 3 inches.

Another question might pertain to the boat’s performance in choppy water conditions. A video shared by Candela on its Facebook page demonstrates its impressive capabilities. It showcases the boat on April 6, paired with the commentary: “From last week’s storm over Stockholm. Rough weather sea trials and comparing the Candela Seven against a 9 meter RIB!”

Unsurprisingly, such remarkable innovation has garnered Candela several prestigious accolades: it was nominated for European Powerboat of the Year, won ‘Best for Future’ at the Best of Boat Awards, and earned the title of ‘Best Foiling Boat’ (being the only electric model) at Foiling Week.

However, the most rewarding recognition may reside in the testimonials from the first buyers highlighted on the Candela website:

“With six people on board, we usually operate at nearly full speed, around 30 knots, and complete the journey from Stockholm to our summer house in just over an hour. Even with all the passengers and gear, we consistently arrive with ample battery capacity remaining.”

Candela’s electric foiling vessel has successfully competed against fossil fuel boats once more, earning the Foiling Motor Boat Award at the recent Foiling Week awards.

Foiling Week, held in Milan this week, celebrates various types of foiling boats and is the only worldwide event focused on “the incredibly fast foiling boats, along with their sailors, designers, and builders.” One of the main sponsors of Foiling Week is Torqeedo, and the award for Candela was presented by Oliver Glück, Torqeedo’s Vice President of Marketing (Left, above).

Candela is certainly in a strong position, having secured the Best of Boat Award in November and also receiving a nomination for the European Powerboat of the Year at BOOT Dusseldorf.

“We’re doing this to lead the green transition at sea,” Gustav Hemming, the city councilor responsible for climate and infrastructure, stated to reporters. The goal of the nine-month pilot project was to “encourage more people to leave their cars behind and opt for a (public transit) card instead.”

Candela’s CEO, Gustav Hasselkrog, was clear in expressing the rationale for shifting away from internal combustion engine (ICE)-powered ferries.

“Traditional ships have seen little evolution over the past century and rank among the least energy-efficient transport options, only competing with a battle tank,” Hasselkrog remarked.

As per Storstockholms Lokaltrafik, boat travel is the fastest-growing mode of public transit in the city, with approximately 6.2 million boat trips recorded in 2022, and the transit agency is considering adding more ferries like the P-12.

The P-12 ferry was “engineered with both passengers and environmental considerations in mind,” Hasselkrog noted, stating that it provides “a highly enjoyable experience without imposing environmental burdens such as wakes, emissions, and noise.”

A life-cycle assessment conducted at the Kungliga Tekniska högskolan, or the Royal Institute of Technology, in Stockholm, titled “Electric Hydrofoil Boats Beat Diesel Boats for Climate Sustainability,” indicates that a Candela P-12 has the potential to produce 97.5% less CO2 throughout its lifespan than a conventional diesel vessel of similar size.

In September, a Candela crew achieved a world record by navigating a hydrofoiling Candela C-8 from Stockholm to the Finnish autonomous region of Åsland. This journey marked the first instance of an electric boat crossing the Baltic Sea.

“Our objective was to showcase that zero-emission marine travel is achievable today, and that foiling electric vessels are significantly more economical to operate than fossil-fueled boats,” Hasselkrog stated.

There were moments of range anxiety during the journey, but it did not stem from the C-8. “The irony is that the photographer’s gasoline-powered chase boat needed to refuel six times during the journey [to Åland and back], while we only needed to charge three times,” Hasselkrog explained.

The round-trip crossing of the Baltic Sea primarily utilized the existing charging infrastructure and received assistance from Empower, a charging solutions provider based in Finland. The voyage began in Frihamn, a Malmö neighborhood, and proceeded to Kapellskär, a port city located 60 miles (90 km) north of Stockholm, where the C-8 was recharged using a 40-kW Kempower wheeled charger linked to the harbor’s power grid. In Mariehamn, the boat was connected to the marina’s three-phase outlet for charging. On the same day at 6 pm, the C-8 team left Åsland, navigating the boat back toward Sweden and stopping again to recharge in Kapellskär. The journey continued despite heavy fog, arriving in Frihamn at 11:30 pm.

In 2014, our founder Gustav gathered a team of specialists in composite engineering, flight control electronics, hydrodynamics, dynamic modeling, and drone control systems. The goal was to discover methods for creating electric boats that could successfully combine both speed and range.

Positioned to transform maritime travel, our hydrofoiling electric boats – the result of over 10,000 hours of committed research and sea trials in Lidingö, Stockholm – effortlessly glide across the surface of the water. By merging state-of-the-art electric propulsion with active stabilization, they set new standards for speed, efficiency, and durability.

Electric ferries are revolutionizing the maritime sector by replacing traditional diesel engines with cleaner, battery-operated options. These vessels promise decreased emissions, lower operating costs, and quieter journeys.

The maritime transport industry, which has historically depended on reliable yet environmentally damaging combustion engine ferries, is at a crucial turning point. As the industry responds to the demand for sustainability, electric vessels are emerging as viable alternatives.

Setting global benchmarks, from the busy ports of Washington State to the beautiful coastlines of Scandinavia, these ferries not only cut emissions but also redefine waterborne public transit for the 21st century.

The transition towards battery-powered ferries is motivated by the complications presented by conventional vessels. Diesel engines, which have long powered ferry services, are infamous for their considerable greenhouse gas emissions, including CO2, methane, and nitrous oxide. These emissions contribute to global warming and local air pollution, with nitrogen and sulfur oxides posing serious risks to human health.

Additionally, the inefficiency of traditional engines results in high fuel consumption and operational expenses. The challenges of emission control add further complexity, often necessitating the use of advanced after-treatment technologies to comply with regulations. Consequently, existing passenger boats face an increasing demand for newer, cleaner, and more economical alternatives that satisfy the requirements of modern transportation and transit systems.

In light of the maritime industry’s challenges, electric propulsion in marine vessels emerges as a promising innovation. The benefits are numerous. Electric ferries incur lower operational and maintenance costs due to simpler motor designs and lesser mechanical wear. These savings create a positive ripple effect across the entire fleet, as electric boats demand less frequent and less expensive maintenance than their combustion engine equivalents.

Another benefit is the quietness that electric propulsion brings to the marine setting. These vessels move through the water almost silently, in stark contrast to the loud diesel engines. This diminishes noise pollution and improves the traveling experience. Furthermore, the built-in redundancy within electric systems provides a layer of reliability, ensuring that boats stay operational even if one part of the system encounters a failure.

Electric ferries offer clear environmental gains, as evident as the clear waters they aim to protect. By transitioning to electric propulsion, passenger boats can considerably lower emissions of harmful gases such as NOx and CO, along with CO2 and soot, making a significant environmental impact.

Take the Puget Sound, where the electric ferry emitted just 25% of the exhaust of its diesel-powered equivalent. In areas with clean electricity grids, electric passenger vessels can reduce greenhouse gas emissions compared to diesel engines significantly. This underscores the substantial environmental advantages these vessels can provide.

Moreover, by minimizing the need for commuter vehicles on shorter routes, electric ferries help to alleviate traffic congestion and its related environmental effects.

While the environmental advantages are a key attraction, the cost-effectiveness and economic benefits of electric ferries are equally persuasive. For instance, an all-electric catamaran operates across the water at a 21% lower energy unit cost than a traditional diesel ferry. This trend isn’t isolated; throughout much of Europe, electric passenger vessels have shown significantly reduced operational costs compared to their diesel counterparts. Although the initial purchase cost of a battery-powered ferry may be higher, the long-term savings in operation and maintenance are evident.

Battery power and energy efficiency in electric ferries form the foundation of their operation, serving as both an energy source and a representation of energy efficiency. Lithium-ion batteries are commonly used due to their capacity to efficiently store and provide large quantities of electricity. Lithium iron phosphate batteries are becoming increasingly popular in maritime uses because of their safety and durability. This indicates a transition towards more robust and dependable power sources for battery-powered boats.

The management of these energy sources is directed by advanced Battery Management Systems (BMS), which guarantee optimal efficiency, safety, and durability of the batteries. Enhancements in quick recharging capabilities are revolutionary. They will allow battery-operated ferries to stay ready for operation and broaden their journeys beyond previously assumed limits.

To grasp the capabilities and limitations of an electric ferry, one must consider its range. Although these ferries generally have a shorter distance capacity than those powered by combustion engines, improvements in battery storage are continuously pushing their limits. Small (though slower) electric ferries, often observed moving through harbors or on short routes, usually cover ranges of 5 to 30 nautical miles. These are backed by battery capacities of 1 to 2 MWh, making them well-suited for frequent docking and charging opportunities.

For medium and larger vessels (also primarily slow-moving), such as those linking islands or serving longer routes, ranges can vary from 20 to over 100 nautical miles. These vessels typically boast battery capacities ranging from 2 MWh to more than 10 MWh.

In contrast, the fast-moving Candela P-12 electric hydrofoil ferry can reach a distance of up to 50 nautical miles at a cruising speed of 25 knots. This highlights the remarkable potential of contemporary ferry technology. By raising its hull out of the water, the P-12 uses its energy for forward propulsion instead of pushing through water, enhancing energy efficiency by over 80% compared to non-foiling vessels.

Despite the increasing drive to electrify ferry fleets, significant challenges persist. A major obstacle is the establishment of reliable charging infrastructure at ports, which is essential for the seamless operation of battery-powered ferries. The charging arrangements can vary from basic household circuits to more intricate fast chargers. Furthermore, access to clean energy and a robust grid is crucial to support the expanding fleet of electric vessels.

Ports and docks often have limited electrical capacity, particularly in isolated island communities. This creates a notable challenge for the broad adoption of electric passenger vessels. Additionally, achieving the right balance between battery size and vessel weight is vital to uphold efficiency and performance. However, potential solutions are emerging, including the development of more robust grids, hybrid systems, and innovative battery technologies. Together, these developments are facilitating a smoother transition to electric fleets. Nonetheless, the more energy-efficient a vessel is, using kilowatt-hour (kWh) batteries instead of megawatt-hour (MWh) batteries, the simpler electrification becomes.

A further substantial challenge is the high initial cost of electric ferries and the necessary infrastructure enhancements. While long-term savings on fuel and maintenance are significant, the upfront expenditure can be intimidating for many operators. Government incentives and subsidies can significantly help mitigate these costs and promote the uptake of electric ferries. Additionally, the maritime transportation sector is witnessing partnerships between the public and private sectors. These collaborations are financing research and development in electric propulsion technologies. Such collaborations are important for accelerating innovation, lowering costs, and creating a sustainable path for ferry fleet electrification.

With every new vessel entering the market, the water transportation sector advances toward a sustainable future. Norway’s MV Ampere, a trailblazer in battery-powered ferries, commenced its journey in 2015. It set a benchmark for following electric ferry initiatives throughout the country. Meanwhile, Wightlink in the United Kingdom is preparing to launch the Solent’s first entirely electric freight and passenger ferry within the next five years.

These advancements signify the latest progress in maritime technology, with ferries like the Candela P-12 Shuttle at the forefront. As these vessels become part of the fleet, they extend the possibilities of ferry services. They also demonstrate a commitment from the maritime industry to shift toward a cleaner, more cost-effective future.

Candela P-12 Shuttle electric hydrofoil ferry

In a time when environmental sustainability and cost-effectiveness are crucial, our Candela P-12 Shuttle stands out as the first electric hydrofoil ferry globally, poised to transform maritime transport.

Traditional ferries find themselves trapped in an unending cycle of inefficiency that adversely affects both operational costs and their environmental footprint. Excessive fuel consumption results in heightened operational expenses and increased ticket prices, discouraging passenger usage and leading to reduced revenue. This situation is worsened by the maintenance costs linked to complicated combustion engines and the ecological damage caused by carbon emissions. Consequently, operators frequently remain caught in a cycle of elevated costs and meager returns, trying to reconcile sustainability and profitability.

The Candela P-12 Shuttle presents an innovative answer to these problems. As an electric hydrofoil ferry, the P-12 Shuttle merges the advantages of electric power with hydrofoil technology. Hydrofoils elevate the hull above the water, which significantly minimizes drag. This enables the P-12 Shuttle to glide effortlessly and efficiently, cutting energy usage by up to 80% compared to conventional ferries. This leads to decreased operational costs and a substantial reduction in greenhouse gas emissions.

The propulsion system of the P-12 completely eliminates reliance on fossil fuels. This drastically lowers fuel expenses and diminishes the carbon footprint associated with ferry operations. The P-12 has the capability to traverse longer distances on a single charge, alleviating range issues that typically constrain electric passenger vessels. Moreover, the reduced drag and diminished wear on components result in lower maintenance costs and an extended operational lifespan. Therefore, the P-12 serves as both an eco-friendly and economically viable option for ferry operators.

Passengers on the P-12 Shuttle can experience a quieter and smoother journey, free from the noise and vibrations typically associated with traditional combustion engines. The innovative design of the ferry also allows for increased speeds and reduced travel times, making it a more appealing choice for both commuters and travelers.

Life Cycle Assessment: Electric hydrofoil boats vs. fossil-fuel alternatives

Life cycle assessments indicate that electric hydrofoil boats have a considerably lower environmental footprint compared to fossil-fuel alternatives. A study by KTH Royal Institute of Technology in Sweden confirms these benefits, showing significant cuts in CO2 emissions. Dennis Olson and Felix Gluunsinger from KTH discovered that the electric hydrofoil leisure craft, Candela C-8, had a markedly lower environmental impact regarding Global Warming Potential and Cumulative Energy Demand compared to its petrol-powered equivalents. They also compared the Candela P-12 with diesel ferries used in Stockholm’s public transport and found that the electric version could decrease environmental impact by 1,670 tons of CO2-equivalent annually.

These results highlight the considerable benefits that electric hydrofoil boats provide over conventional marine vessels. By focusing on electric propulsion, marine operators can greatly lessen their environmental impact. This strategy paves the way toward a cleaner and greener future for maritime transportation.

The future of electric ferries: Innovations and prospects

The maritime sector is navigating towards a future defined by innovation and sustainability. The potential for electric ferries is vast. For instance, Stena Line’s next-generation E-flexer vessels will be dual-fuel methanol hybrids, demonstrating the industry’s flexibility and dedication to lowering emissions. Concurrently, San Francisco is set to introduce the country’s first high-speed, high-capacity zero-emission ferry service, establishing a new benchmark for urban transport.

The creation of hybrid vessels by companies such as Brittany Ferries and Isle of Man Steam Packet Company showcases various methods of electrification. These advancements indicate a future in which battery-powered ferries will be crucial in the global movement for zero-emission maritime operations.

Conclusion

Electric ferries mark a significant transformation in maritime transport, resulting in lower environmental impacts and operational expenses. With continued progress in battery technology and charging infrastructure, these vessels are expected to become more feasible, making them an attractive option for both ferry operators and passengers.

Frequently Asked Questions

Are electric ferries more costly to operate than diesel ferries?

No, they are usually less expensive to operate due to lower energy consumption, decreased maintenance needs, and possible government subsidies.

How do electric ferries affect local communities and ports?

Electric ferries can benefit local communities and ports by diminishing air and noise pollution, enhancing quality of life, and improving the passenger experience.

What type of battery technology is utilized in electric ferries?

Electric ferries typically employ lithium-ion and lithium iron phosphate batteries for their high energy density, efficiency, and safety, managed by advanced systems for peak performance.

What developments are underway to expand the range of electric ferries?

Improvements in battery storage, efficiency, and rapid charging facilities at ports are enhancing the operational range of electric ferries. Additionally, hybrid systems with supplementary power sources are being examined for increased range and flexibility.

How do electric ferries contribute to lowering greenhouse gas emissions?

The environmental benefit is largely contingent upon how renewable the electricity grid is. Sweden relies heavily on renewables for electricity, though this isn’t universally true. Regardless, electric ferries produce no exhaust emissions locally, thus eliminating harmful outputs typically related to diesel engines.

Can electric ferries cover as much distance as diesel-powered ferries before needing to recharge?

Though electric ferries generally have a shorter range compared to diesel ferries, advancements in battery technology are gradually extending travel distances, and fast-charging infrastructure is being developed for quicker charging.

What are the primary obstacles to the widespread adoption of electric ferries?

High initial costs and the lack of fast-charging infrastructure represent significant hurdles for the expansion of the electric ferry market. Additionally, range limitations compared to traditional ferries pose challenges for commercial viability.

What are the key factors propelling the adoption of electric ferries?

Government initiatives and subsidies are critical drivers behind the adoption of electric ferries, facilitating the shift towards more sustainable maritime transportation solutions.

November 17, 2024
Hybrid vs. electric cars: What’s the difference?
For a lot of car purchasers, hybrids appear to be a good middle ground between gasoline and electric vehicles. Hybrids, which combine gasoline engines with electric motors, generate less pollution and consume less fuel compared to traditional cars. Additionally, drivers never need to fret about depleting battery power on a deserted highway.

However, while hybrids can save some individuals money, that isn’t universally true. Numerous experts and environmental organizations criticize hybrids, arguing that the financial savings are overstated and that they do not sufficiently reduce greenhouse gas emissions to help mitigate global warming.

Nonetheless, many car buyers seem convinced of their advantages. Hybrid car sales in the U.S. rose by 33 percent from January to July compared to last year, representing 11 percent of new car sales, according to government figures.

Here’s what to consider if you’re shopping for a car.
- Not all hybrids are the same
- There are two primary types of hybrids: conventional and plug-in. Conventional hybrids include systems that recover some energy from braking to charge a battery. The stored energy then powers an electric motor that supports the gasoline engine.
This arrangement helps to offset the inherent inefficiency of gasoline engines. A gasoline vehicle generally converts less than one-third of the energy in a gallon of gas into movement. Much of the energy is lost through the brakes, which transform motion into heat.

By recapturing some of that wasted energy to recharge a battery, the base model of the Toyota Prius, for example, achieves an estimated 57 miles per gallon according to the Environmental Protection Agency.

Nevertheless, conventional hybrids still consume gasoline. This means they contribute to climate change and produce harmful air pollutants. Essentially, they are just more efficient gasoline-powered vehicles.

‘Go a month’ without filling up

Plug-in hybrids boast larger batteries that allow cars to run solely on electricity for limited distances. They can be plugged into the same charging stations as electric vehicles and standard electrical outlets.

The plug-in version of the Toyota RAV4 SUV can travel around 40 miles on battery power alone, which exceeds the average daily distance most Americans drive. A gasoline engine engages when the battery is exhausted. In theory, owners of plug-in hybrids don’t need to refuel unless they are embarking on a long road trip.

“You could go a month without putting any gas in your vehicle,” stated Jack Hollis, the chief operating officer of Toyota Motor North America. “On your weekends,” he added, “when you’re looking to escape and want to travel two, three, or four hundred miles, having that plug-in hybrid is really beneficial.”

Mr. Hollis also pointed out that hybrids are typically more budget-friendly than entirely electric vehicles. They often cost only a few hundred dollars more than their gasoline counterparts. Electric vehicles generally come with a price tag that is several thousand dollars higher.

Several plug-in hybrids, such as certain models of the Jeep Wrangler and Ford Escape, are eligible for federal tax credits of up to $3,750, which contributes to possible savings on fuel expenditures. (No Toyotas qualify because they have too many imported parts.)

Superior to fully electric?

While hybrids are less harmful to the environment than gasoline vehicles, fully electric cars serve as a much more potent ally in the fight against climate change, according to Peter Slowik, who specializes in passenger cars at the International Council on Clean Transportation, a nonprofit research and advocacy organization.

“There is no realistic path to achieving our climate objectives with any vehicles that burn fossil fuels,” Mr. Slowik commented.

A plug-in hybrid emits twice the greenhouse gas emissions of a fully electric vehicle over its lifetime, according to research by the council that considers emissions from production and from recycling batteries once a vehicle reaches the end of its lifespan.

Another potential disadvantage of plug-in hybrids is their complexity, which can increase maintenance expenses. Plug-in hybrids comprise more moving parts that may malfunction. Unlike electric vehicles, both types of hybrids require regular oil changes. Generally, both hybrids and electric vehicles tend to have higher insurance costs than gasoline-powered cars.

That said, hybrids are generally dependable because they typically come from manufacturers like Toyota or Honda, known for their reliable products, as pointed out by Keith Barry, a senior writer at Consumer Reports who covers the automotive sector. “The manufacturers that make hybrids tend to be the most reliable overall,” he noted.

It’s crucial to understand that plug-in hybrids provide minimal benefits if their owners neglect to keep them charged. According to a report by the International Council on Clean Transportation, many users do not charge them regularly.

Owners of plug-in hybrids with a range of 40 miles typically utilized only 45 percent battery power, significantly less than the assumptions made by the Environmental Protection Agency when estimating fuel economy. The study, which examined millions of miles of vehicle data, showed that real fuel consumption could be up to two-thirds higher than the E.P.A. predictions.

To sum up, Mr. Barry emphasized that plug-in hybrids can be both environmentally and financially advantageous for individuals who can charge at home or at work. The decision hinges on factors such as driving frequency, electricity costs, and the price range being considered.

If you’re considering a smaller vehicle, “a hybrid is likely the better choice,” he noted. This is because there are still relatively few affordable small electric vehicle options available. However, Mr. Barry mentioned, “if you’re exploring a luxury or sporty car, typically, the electric version will save you some money.”

The case for hybrids is likely to weaken in the coming years. The cost of fully electric vehicles is dropping quickly, and there is an increasing variety of models available. Public charging stations are becoming more widespread, advancements in technology are decreasing charging times, and vehicles capable of traveling over 300 miles on a single charge are becoming the norm.

The fully electric variant of General Motors’ Chevrolet Equinox, which has just begun to reach dealerships, starts at $27,500 after factoring in a federal tax credit. The Equinox, which can exceed 300 miles on one charge, is the first in a series of affordable electric vehicles expected in the coming years, potentially making hybrids and gasoline vehicles less appealing price-wise, not to mention the fuel savings. In contrast, a Toyota RAV4 plug-in begins at $43,700.

“For some people, plug-in hybrids can serve as a good transitional option, especially for those traveling long distances without charging access,” said Katherine Garcia, director of the Clean Transportation for All Campaign at the Sierra Club. “However, our goal is to inform people that, given the climate crisis, we really need to move away from fossil fuels.”

Various new and expanding options are available at car dealerships regarding hybrid and electric vehicles, but recent decisions from major automakers indicate a shift towards more hybrids rather than EVs. Ford recently announced a delay in its electric pickup and is currently focusing more on its North American hybrid lineup.

The notion of “EV enthusiasm has waned,” with the idea of “consumer choice” becoming prominent again among automakers like Ford, General Motors, Mercedes-Benz, Volkswagen, Jaguar Land Rover, and Aston Martin, all of which are revising or postponing their electric vehicle plans. GM’s EV sales remained minimal during the most recent quarter.

A Gallup poll released on Monday found that only 44% of U.S. adults are “seriously considering or might consider” purchasing an EV, down from 55% in 2023. The percentage of people not intending to buy an EV has risen from 41% to 48%.

However, determining the best value can be quite complex. These choices often depend on various elements such as initial cost, driving patterns, how long you intend to keep the car, anticipated ongoing expenses, and even your geographical location.

The solution isn’t always clear, even with headlines favoring hybrids. Here are some insights to assist car buyers in making the right choice.

Assess your driving habits

Before you begin comparing expenses, it is logical to evaluate how you will use the vehicle.

Are you merely commuting five or ten miles daily for work, or are you planning long road trips? If you frequently drive long distances, consider the availability of fast-charging stations along your route. If fast-charging stations are rare, as they are in many regions, a hybrid might be more beneficial where you can conveniently refuel at a gas station, according to Sandeep Rao, lead researcher for Leverage Shares, which provides investment funds focused on stocks of both EV and traditional automakers.

The federal initiative aimed at developing a comprehensive charging network across the U.S. has yet to be fully realized. Currently, efforts have concentrated on specific regions like California, the New York tri-state area, Florida, and Texas, while the vast majority of people live in the areas in between. “Most Americans lack access to EVs due to insufficient charging infrastructure,” Rao explained.

He also recommended considering how long you plan to own the vehicle, potential maintenance needs, and what nearby service options are available. Other elements to take into account include your home setup. Do you possess the appropriate conditions for convenient and quick EV charging? What are the upfront costs for upgrading your system to enable faster charging?

Evaluate the initial investment, EV vs. hybrid

If you’re still uncertain between an EV and a hybrid, the next step is to evaluate upfront expenses.

The mean price for the top ten best-selling electric vehicles in the U.S. stands at approximately $53,758, with an average of $48,430 for the entry-level versions of each model and $64,936 for the premium versions, as reported by Find My Electric, an independent EV marketplace. These ten EVs have price points that span from $26,599 for the Chevrolet Bolt EV to $99,000 for the highest-priced variant of the Rivian R1S.

In comparison, the average starting price for hybrid vehicles is $33,214 according to iSeeCars.com, a vehicle search platform. If you have particular models in mind, the Department of Energy provides a tool that allows you to compare up to four vehicles at once. You can also evaluate different models based on their fuel efficiency.

Look into potential auto rebates and incentives

If you’re favoring an EV but are still concerned about the initial price, explore available rebates. There are federal subsidies available — up to a maximum of $7,500 — but qualifying for these is becoming increasingly challenging as more manufacturers become ineligible.

Additionally, investigate state and local incentives. Buyers can check the Electric for All website, curated by Veloz, a nonprofit organization, to find incentives like vehicle tax credits and rebates, charging rebates, local utility incentive programs, and other special benefits for choosing electric.

“Depending on your location, you may be able to purchase an EV at a price point comparable to that of a hybrid or internal combustion vehicle,” stated Steve Christensen, executive director of the Responsible Battery Coalition, a nonprofit group dedicated to the responsible management of batteries.

Consider a plug-in hybrid

Another option worth considering is a plug-in hybrid electric vehicle, which can be an appealing choice for individuals transitioning from gas or diesel vehicles to battery-powered options.

The primary distinctions between full hybrids and plug-in hybrids involve the size, cost, and functionality of their electric batteries, as explained in an online Q&A from Progressive Casualty Insurance Company. Moreover, a plug-in hybrid can be charged at home or at public charging stations, while a full hybrid recharges its battery using its gasoline engine.

If you’re contemplating a plug-in hybrid, the Department of Energy offers a calculator that can help you estimate personalized fuel consumption and expenses based on your driving habits, fuel prices, and charging times.

Emphasize the total cost of ownership, not just initial expenses

Typically, the initial costs associated with EVs tend to be higher; however, it might be more advantageous in the long run.

For instance, smaller EVs, such as compact cars or sedans with a range of roughly 200 miles, can reach a break-even point with similarly sized traditional hybrids in five years or less, according to a recent study from the University of Michigan. Notably, this is without considering any incentives, as noted by Maxwell Woody, a PhD candidate at the University of Michigan and lead author of the research.

On the other hand, larger vehicles, including midsize SUVs and pickup trucks with an extended range of up to 400 miles, do not achieve break-even points with hybrids, even with incentives factored in, according to the study. It’s significant to mention that this data is based on historical battery prices, which have seen a significant decline in recent years and are projected to continue decreasing, suggesting that electric vehicles will generally perform better soon, Woody stated.

Calculating expenses for a plug-in hybrid is more complex since operational costs can vary significantly based on how often you charge versus fill up with gas. If you use it exclusively on electricity for urban travel, for instance, your costs could closely resemble those of an EV, Woody remarked. In contrast, during long trips, the expenses for refueling might align more closely with those of a gasoline vehicle, he added.

When assessing the overall cost of ownership, it’s essential to account for maintenance expenses, advised Albert Gore, executive director of ZETA, an industry-supported coalition advocating for full EV adoption. He references a study from Argonne National Lab indicating that maintenance costs per mile are considerably lower for an EV compared to traditional hybrids or plug-in hybrids.

Additionally, ensure you are making direct comparisons in terms of features, model, year, quality, and intended use cases, Woody emphasized. For instance, someone evaluating a Nissan Leaf, which is fully electric, might consider comparable data for a Honda Civic hybrid, he noted.

What distinguishes hybrid cars from fully electric cars?

As fuel prices seem to keep rising, having a fuel-efficient vehicle has become essential for many people in New Zealand. Therefore, it’s logical to explore options like hybrids and electric vehicles, alongside more traditional choices. To ensure that you choose the vehicle that best suits your family’s needs, it’s important to grasp the distinctions between hybrids and battery electric vehicles.

Hybrid cars utilize an electric motor to accelerate and travel at speeds up to about 25 km/h. As your speed increases, the petrol engine activates, and when you decelerate or brake, the energy is captured in the battery for future use. Hybrid vehicles do not require plugging in to recharge the battery.

In contrast, electric cars (BEVs) operate without petrol, produce no exhaust emissions, lack a clutch or gears, have fewer moving components, need less maintenance, and are extremely quiet. They function entirely on one or more electric motors and must be charged by plugging them in, either at home or at a public charging station. The downside is that they usually offer a shorter driving range compared to hybrid cars; however, it’s easy to stop and recharge the battery during your drive.

There is also a third option: plug-in hybrid electric vehicles (PHEVs), which are hybrids that can also be recharged by plugging in. These vehicles provide more electric driving range than standard hybrids, as well as the convenience of a petrol engine that activates when necessary or once the electric battery is drained.

Who are hybrid cars most suitable for?

Hybrid vehicles are highly fuel-efficient and inexpensive to operate compared to their petrol-fueled counterparts, which is why cars like the Toyota Corolla Hybrid and Toyota Camry Hybrid are very popular among urban dwellers. The ongoing low-speed, stop-and-go driving conditions typical of city commuting are where hybrids excel.

When driving at slower speeds or crawling in heavy traffic, hybrids utilize electric power, avoiding the consumption of petrol. Hybrid vehicles also perform well on highways or motorways, as they often incorporate Atkinson cycle engines supported by electric motors.

Hybrids are ideal for individuals seeking a dependable vehicle that generates less pollution than petrol or diesel options and who frequently drive in urban environments. They cater to those who prefer not to plug in their vehicle or experience range anxiety. Additionally, they can be more affordable than PHEVs or BEVs.

Who are plug-in hybrid electric vehicles (PHEVs) best suited for?

PHEVs occupy a middle ground between hybrids and EVs, providing drivers with the advantages of both types. Equipped with both electric motors and a petrol engine, they can also be plugged in. They are suitable for individuals who want to minimize fuel use on short trips but still desire the option of a traditional vehicle for longer journeys.

While you will eventually need to refuel, PHEVs like the Prius Prime average around 1.0 litre per 100km*, which is very cost-effective.

*Fuel consumption figures are assessed under controlled conditions and provided for comparison purposes; actual results may vary based on vehicle usage and operating circumstances.

Who are battery electric cars best intended for?

Electric vehicles may cost more than hybrids, and unlike PHEVs, EVs don’t automatically transition from electric mode to hybrid mode when the battery runs out. Newer EVs typically offer a range of approximately 200-300 km, while older models have a range closer to 100 km.

Electric cars are suited for those who primarily make short trips and appreciate the concept of zero-emission driving. For longer journeys, it’s essential to ensure that charging stations are available along the route and to allow ample time for recharging. However, this concern is diminishing with the advent of new EVs that can achieve ranges of up to 500 km.

A Comparative Analysis

Are you trying to reduce your car’s fuel expenses? Electric and hybrid vehicles are excellent options to consider.

Now is an ideal time to transition; models of hybrids and electric cars from 2023 and 2024 provide outstanding fuel efficiency, along with potential incentives and rebates to lower the price.

In this article, we will conduct a direct comparison between hybrid and electric vehicles to determine which one offers the most savings and serves as the superior choice overall.

Hybrid vs. Electric Vehicles Overview
- Hybrid cars utilize gasoline engines supplemented by small electric motors.
- Electric vehicles operate entirely on large battery-driven electric motors.
- Hybrids are more affordable initially but qualify for fewer incentives and rebates compared to electric cars.
- When compared to gasoline prices, electric vehicles can be up to 70% cheaper, while hybrids can be up to 60% cheaper at best.
- Electric vehicles are significantly less expensive to maintain than hybrids due to having far fewer moving parts.
- While electric cars have a higher initial cost, they often result in lower lifetime expenses due to fuel savings, incentives, and lower maintenance costs, although individual experiences may vary.
Before we proceed with the cost analysis, here’s a brief overview of how hybrid and electric vehicles function.

Electric cars exclusively use powerful electric motors connected to large rechargeable batteries. They are often referred to as battery electric vehicles (BEVs) or simply electric vehicles (EVs). Popular examples of electric vehicles include models from Tesla and the Hyundai Ioniq 5.

Hybrid vehicles combine conventional gasoline cars with pure electric cars. Each hybrid features a standard combustion engine (running on gasoline or diesel) alongside one or more small electric motors powered by a battery. Because the gas engines do most of the work, they are frequently termed “part-time electric cars.”

There are two primary subcategories of hybrids to note:
- Plug-in hybrid electric vehicles (PHEVs): These hybrids recharge their batteries by plugging into a charging station or wall outlet. They usually can operate for short trips (20 to 30 miles) using electric power before transitioning to a gas-fueled hybrid mode. Notable PHEVs include the Jeep Wrangler 4xe and the Toyota Prius Prime.
- Regular hybrids, or HEVs: These vehicles replenish their battery packs using the gasoline engine and regenerative braking. The smaller electric motors assist the gas engine, either extending the driving range or enhancing performance. The Honda Insight is a well-known example of a standard hybrid seen on the roads.
Which is More Affordable Upfront?
Winner: Hybrids

Hybrids are generally cost-effective, with many models priced between $25,000 and $35,000. The Honda Insight, a standard hybrid, starts at roughly $25,000, while the Toyota Prius Prime begins around $28,000.

Electric vehicles often have a higher price point, particularly if you seek one with a longer range. For instance, Tesla models exceeding 300 miles of range all start at prices above $60,000. The all-electric Chevy Bolt is a more affordable option, with starting prices in the mid $30,000s, but it only offers a range of 259 miles. This is because having a longer range necessitates a larger, more powerful battery, and battery prices remain high.

Clearly, hybrids win in this category.

Which Offers Better Incentives and Rebates?
Winner: Electric Cars

Electric vehicles stand out significantly when it comes to incentives and rebates. There are some available for plug-in hybrid vehicles, but options for regular hybrids are limited.

The most common incentive is the federal EV tax credit, which can provide up to $7,500 off the purchase of a new vehicle. This credit is available for qualifying electric vehicles and plug-in hybrids but does not apply to regular hybrids.

You may also discover additional incentives provided by your state, local authorities, or utility companies; numerous programs exist across the country. For example, New Jersey’s Charge Up NJ initiative promotes the purchase or lease of fuel-efficient vehicles priced under $55,000, offering rebates of up to $4,000 for eligible electric vehicles and $1,050 for plug-in hybrids. Unfortunately, regular hybrids are not eligible for rebates.

New Jersey’s Charge Up NJ program exemplifies incentive schemes typically offered at the local level. When we analyzed a random selection of 20 such programs, we observed:
- All of them included battery electric vehicles (BEVs).
- Most (but not all) covered plug-in hybrids (PHEVs), though the incentive amounts are usually lower.
- None included regular hybrids.
Which has lower fuel expenses?
Winner: Electric vehicles

Both electric vehicles and hybrids are considerably cheaper to operate than traditional gas cars, but electric vehicles have a slight advantage.

Charging a Tesla with grid electricity costs approximately 4 to 5 cents per mile. This is 70-75% less than the typical fuel cost of an average gas vehicle, which is around 16 cents per mile.

In the case of a highly efficient hybrid like the Kia Niro Hybrid, the average fuel costs (combining gas and electricity) are a bit higher, falling within the range of 6 to 8 cents per mile. This is roughly 50 to 60% cheaper than a conventional gas vehicle.

For a highly-efficient gas hybrid such as the Honda Insight, the fuel expenses amount to about 7 cents per mile, which equates to approximately 55% less than a regular gas car.

In summary, electric vehicles are up to 70% more economical than gas cars, while hybrids are maximally 60% more affordable. Electric vehicles take the win by a narrow margin.

Which is more affordable to upkeep?
Winner: Electric vehicles

Electric vehicles are significantly cheaper to maintain compared to any other type of vehicle.

EVs require less maintenance due to not having a conventional engine and its many components. This means no more oil changes, and you won’t have to worry about replacing gaskets, cylinder heads, spark plugs, and so forth. You also don’t need emissions testing because they don’t have exhaust systems.

Overall, owning an electric vehicle can cost $400 to $1,000 less in maintenance each year compared to gasoline cars.

On the other hand, hybrid vehicles are not less expensive to maintain than gas cars; in fact, they might be more costly. Hybrids include all the moving parts typical of gas cars, plus additional components necessary for the electric system.

Hybrids vs. electric vehicles: Overall cost assessment

Electric vehicles are the evident champions. They qualify for more incentives and rebates, and their operational costs are significantly lower due to reduced ‘fuel’ and maintenance expenses.

Savings from electric vehicles accumulate over time, so they provide the most benefit if you plan to drive considerable distances or keep your vehicle for many years.

The main advantage of hybrid cars is their lower initial purchase price, but this benefit is often diminished—or even negated—by the substantial incentives and rebates available for electric vehicles.

Still undecided on which to select? You might want to set aside the calculator and think about other non-monetary factors. After all, financial considerations aren’t everything! If you prefer not to think about charging on a long road trip, opt for a hybrid. If you’re thrilled by immediate torque and a rapid 0-60 acceleration, fully commit to an electric vehicle! No matter which route you choose, both options will save you money compared to a traditional gas vehicle.

Utilize solar panels for economical charging

With a gasoline vehicle, control over your fuel expenses is limited—you must pay the rates determined by local gas stations.

However, with an electric vehicle or plug-in hybrid, you gain significantly more control over your charging costs. You can charge at public stations, at home using grid electricity, or with solar panels at your residence.

Here’s a breakdown of costs associated with each charging option:
- Public charging: $0.28 – $0.69 per kWh
- Grid power at home: $0.10 – $0.40 per kWh
- Home solar panels: $0.05 – $0.11 per kWh
Charging an EV’s battery using home solar panels typically proves to be the most economical method, averaging around $0.11 per kilowatt-hour. In contrast, using power from your utility supplier is likely to cost closer to $0.15 per kilowatt-hour. Additionally, a significant advantage of solar panels is their ability to power your entire home, substantially lowering your household energy expenses.

Key distinctions between hybrids and electric vehicles (EVs)

If you’re interested in using electricity to power your driving, there are currently three choices that permit either all-electric or a combination of petrol or diesel fuel and electric power.

The first is the battery electric vehicle (BEV), which is entirely driven by a motor running on electricity. Electric vehicles must be connected to a power source to replenish the battery that propels the car.

The mild hybrid or plug-in hybrid model utilizes both electricity and petrol or diesel to operate a vehicle. A plug-in hybrid vehicle (PHEV) comprises a combustion engine and an electric motor, accompanied by a smaller battery than that found in an EV.

Like an EV, a PHEV’s battery also needs to be plugged in to recharge. However, its range is shorter than that of fully electric cars, usually extending up to 50 miles. This makes plug-in hybrids more suitable for shorter trips.

Lastly, the system that relies the least on battery power is the full hybrid. A fully hybrid electric vehicle combines a combustion engine with an electric motor to create motion.

The main distinction from a PHEV is the even smaller battery that cannot be charged by plugging it in. Instead, it is powered by energy produced from the combustion engine and regeneration from braking, which may be referred to as regenerative braking.

Fully electric
Fully electric vehicles, sometimes known as battery electric vehicles (BEVs), rely on electricity stored in a battery for propulsion. EVs require charging, which can be completed at home using a charge point like our Solo range or a standard 3-pin plug, or at public charging stations found at workplaces or supermarket parking lots.

Over the years, the average distance that these vehicles can travel on a single charge has more than doubled, increasing from about 100 miles in 2011 to 250 miles in 2024. Some high-end models can even provide a range of over 400 miles on a single charge.

The popularity of fully electric vehicles continues to rise, with 315,000 EVs registered in 2023, representing an 18% increase compared to the previous year.

Advantages of a battery electric vehicle

The advantages of BEVs can be summarized as follows:
- Incentives
- No road tax
- Lower operational costs
- Improved air quality
- Distinct driving experience
Incentives available for BEVs

If you own or use a BEV via leasing or have access in another manner, you could receive up to £350 off the price of both purchasing and installing a home charger.

This incentive is beneficial for EV owners, enhancing the convenience of home charging, which is more accessible than using a 3-pin plug socket and faster.

The electric vehicle charge point grant is available for those living in flats or renting homes with private off-street parking. Additional criteria must be met to qualify, which you can check on the Government’s website.

You can explore the comprehensive list of available commercial and private grants in our guide to Government grants for electric vehicles.

No road tax for BEVs

All battery electric vehicles currently incur £0 in road tax.

However, from 1 April 2025, road tax regulations will alter. The existing £0 per year band A will be abolished, and vehicles in this band will transition to band B, which costs £20 annually. EVs registered on or after April 2017 will be subject to the new Standard Rate of £180 annually.

Although by April 2025 BEV drivers will lose the advantage of this tax exemption, the lower costs associated with charging an electric vehicle will still contribute to a reduction in the overall expense of owning and driving an EV.

You can find detailed information about road tax regulations and other relevant details in our extensive EV buying guide.

Lower expenses for BEVs

Operating a BEV is less costly compared to running a petrol or diesel vehicle.

You can benefit from competitive electricity rates when charging at home, including options that offer lower prices during off-peak hours.

Additionally, EVs are cheaper to service and maintain due to their electric motors and batteries, which feature fewer mechanical parts that may require frequent repairs or replacement.

We discuss expected costs related to owning and maintaining a battery electric vehicle in our EV buying guide.

Improved air quality thanks to BEVs

Only battery electric vehicles produce zero tailpipe emissions. Therefore, they do not release harmful pollutants while the electric motor and battery are in operation, thus contributing to better air quality.

Cars with internal combustion engines (ICE), along with plug-in hybrids and full hybrids, do emit pollutants that worsen air quality. The substances and chemicals responsible for pollution are associated with causing and exacerbating various health problems, including asthma, pneumonia, and lung cancer.

Unique driving experience

Driving BEVs provides a somewhat distinctive experience. Because power is delivered instantly from the battery to the electric motor, EVs accelerate quickly.

Traveling in an EV is generally a much quieter and smoother experience compared to an ICE vehicle. The absence of a combustion engine, gears, and other moving parts results in less vibration and noise.

Moreover, many contemporary EVs are equipped with excellent features, such as advanced infotainment systems, integrated cameras, and assistance systems.

Disadvantages of Battery Electric Vehicles (BEVs)

The drawbacks of BEVs include:
- Discrepancy between official and actual range
- Purchase cost
- Charging challenges
- Rapid advancements
Official range vs. real-world range of BEVs

You might discover that the official range provided by a BEV manufacturer does not align with the range you experience in reality. Manufacturers utilize the Worldwide Harmonized Light Vehicles Test Procedure (WLTP) to estimate a vehicle’s range.

Although the WLTP range is more reliable than prior methods due to its incorporation of real-world driving conditions, the actual range can vary based on your driving habits, temperature conditions, and other variables.

Cost of acquiring a BEV

Generally, battery electric vehicles are pricier than their petrol or diesel counterparts.

If you drive frequently and prefer ownership, investing in an EV can be financially beneficial in the long run. Reduced expenses for charging, servicing, and maintenance will result in a lower overall ownership cost over time.

However, if you enjoy changing vehicles every few years, leasing might offer a more cost-effective route to driving an EV. There are competitive leasing deals available that can make leasing an EV comparable to owning a similar internal combustion engine vehicle.

Rapid technological advancements

The rate at which the electric vehicle industry is innovating is extremely high. While this is positive for widespread adoption as it makes EVs more appealing to a broader audience, it also implies that a vehicle that is cutting-edge today may seem outdated in just a matter of years.

If you wish to keep up with the latest advancements and benefit from continual innovations, consider flexible ownership options like leasing.

Hybrid Vehicles (Mild and Full)

Hybrid electric vehicles incorporate both a combustion engine and an electric motor, allowing them to utilize both sources simultaneously.

A full hybrid relies primarily on petrol or diesel for power but includes a small battery that recovers energy from braking and the combustion engine for recharging. This type cannot be plugged in for battery charging, so it remains dependent on fuel.

Mild hybrids, also known as plug-in hybrid electric vehicles (PHEVs), operate on the same principle of combining electric and fuel power for propulsion. The primary distinction is that a plug-in hybrid can recharge its battery through a charging station or a standard 3-pin plug.

Compared to fully battery electric vehicles, mild and full hybrids provide a relatively limited electric range.

Advantages of Hybrids

The benefits of hybrid vehicles consist of:

1. Familiar driving experience
2. Enhanced efficiency
3. Lower road tax
4. Towing capabilities
5. Extended range

Familiarity with hybrids

If you’ve previously owned an internal combustion engine (ICE) vehicle, driving a full or mild hybrid will likely feel quite similar. You will still need to visit a petrol station to refuel and won’t depend solely on a charged battery to operate.

Enhanced efficiency through hybrid technology

Hybrids provide improved efficiency by effectively combining electric and fuel power to drive the vehicle.

Utilizing regenerative braking is crucial in minimizing fuel consumption, as it captures energy that would otherwise be wasted during braking. This feature renders hybrid vehicles more efficient compared to traditional petrol or diesel cars, especially in urban environments with frequent stop-and-go traffic.

Lower road tax for hybrids

Hybrid vehicles may qualify for a reduced road tax rate, though it is not eliminated entirely like it is for electric vehicles.

Hybrids registered on or after April 1 20217 will incur a road tax fee ranging from £0 to £120 in their first year, followed by £170 each subsequent year. This is £10 less than what an equivalent petrol or diesel vehicle would owe.

Additional charges apply to hybrids with a list price exceeding £40,000. For more details, refer to our road tax guide.

Towing capabilities

For those who frequently tow a caravan or trailer, opting for a full or mild hybrid might be the most suitable choice. It offers some electric driving advantages while retaining the ability to tow heavier loads.

Many electric vehicles are not permitted for towing, and if they are, towing will impact their range as more energy is required to pull additional weight, reducing the driving distance. Although a hybrid vehicle may not match the towing capacity of an ICE vehicle, it certainly exceeds what an electric vehicle can tow.

Extended range with combustion fuel

During long trips, a hybrid may achieve a greater distance than an electric vehicle, although advancements in battery technology are steadily increasing EV ranges. The superior fuel efficiency and predominant use of a combustion engine render hybrid vehicles more dependable for long-distance travel, plus you’ll need fewer stops for refueling.

Electric range of hybridsThe electric range for hybrids is quite limited. In a full hybrid, you can travel only 1 to 2 miles using the electric motor, while a mild hybrid can go up to 50 miles.

A Plug-in Hybrid Electric Vehicle (PHEV) is designed to handle most short trips using just battery power, comfortably covering the average 8-mile journey taken by UK drivers.

Cost of buying a hybrid

Typically, hybrids come at a higher purchase price compared to comparable petrol or diesel vehicles. They are also less economical for motorway driving than diesel cars due to their inefficiency in that environment.

For instance, the new petrol version of the Vauxhall Astra starts at a price of £26,960, whereas the PHEV variant begins at £37,935. This means the PHEV costs about £11,000 more than the traditional combustion engine version and even more than the fully electric Astra.

Upcoming ban of hybrid cars

Starting in 2035, the sale of new full and mild hybrid vehicles will be prohibited. This indicates that you can no longer purchase brand new hybrids, although buying and selling used ones will still be permitted.

This legislation may decrease the attractiveness of hybrids to consumers and could influence their resale values.

Adapting to new charging habits

If you’re transitioning from an internal combustion engine (ICE) vehicle, you will need to become accustomed to charging your mild hybrid in addition to refueling. While this adjustment might seem minor, it’s significant because a PHEV operates most efficiently when the battery is utilized.

Failing to charge the battery and depending solely on the combustion engine leads to lower fuel efficiency since you are carrying a heavy battery without taking advantage of its benefits.

Driving experience in a hybrid

The extra weight of the battery can influence the driving experience. Riding in a hybrid is not uncomfortable, but it may feel less smooth on uneven surfaces compared to an ICE or electric vehicle, as the added battery weight necessitates a stiffer suspension. You could also find that cornering is more challenging than in other cars.

So should you get a hybrid or an EV?

In the end, we believe that fully electric vehicles are the superior choice if you’re deciding between a hybrid and a battery electric vehicle. With zero emissions, they present a significantly better environmental option for EVs, are more cost-effective to operate, and provide an enjoyable driving experience.

If you’re considering acquiring an EV, take a look at our EV buying guide for tips on purchasing options, incentives, and maintenance.

The hybrid vehicle, which combines a gasoline engine with an electric motor, is gaining attention after being eclipsed by striking electric vehicles from brands like Tesla.

Sales of electric vehicles (EVs) have decreased in the U.S., as reported by car analytics site Edmunds, with the average time for a car to sell after arriving at a dealership increasing from 25 days at the start of 2023 to 72 days a year later. This measure, known as “days to turn,” is an effective indicator of consumer demand.

The significant tripling in the days to turn for EVs is remarkable and not aligned with trends seen in other vehicle categories. Conventional internal combustion engines saw their days to turn increase from 34 to 52 during the same timeframe, according to Edmunds data.

In contrast, standard hybrids outperformed the other categories in terms of popularity, with their days on the lot rising from 16 to 25 over the same period, according to data from Edmunds.

According to Morgan Stanley, hybrid sales grew five times quicker than EV sales in February 2024.

When the Toyota Prius debuted in the U.S. in 2000, it unexpectedly became a popular choice among celebrities and frugal consumers seeking to cut gas expenses. Its lack of flashiness or luxury made its broad appeal even more surprising.

However, this changed when Tesla ignited interest in electric vehicles with its stylish, speedy Roadster and Model S, pushing hybrids into the background. Major automakers quickly followed suit, aiming to join the EV trend — with the exception of Toyota, which lagged behind its competitors. To date, Toyota, the world’s largest car manufacturer, offers only two EV models: the bZ4X and the Lexus RZ, neither of which are sold in significant quantities.

Advocates for electric vehicles and environmental organizations have claimed that Toyota has been working to hinder the EV revolution it has not fully embraced. Although they have previously explored options for battery-electric vehicles like the RAV EV, the company has consistently maintained that the transition to complete electrification will be prolonged and that many consumers are not yet prepared for fully electric vehicles.

Yet, in late 2021, Toyota announced plans to launch 30 new EV models by 2030, with an aim for annual sales of 3.5 million vehicles.

Nearly two years later, hybrid and plug-in sales grew by almost 28% compared to the previous year, representing 30% of the Japanese giant’s portfolio.

Toyota is not the only manufacturer taking advantage of the rising hybrid market.

Hyundai may introduce hybrids at a factory in Georgia that was initially planned to focus solely on electric vehicles. Last year, Ford announced it would reduce output of certain EV models, including the F-150 Lightning electric pickup, in favor of producing more hybrids. General Motors, whose CEO, Mary Barra, has consistently stated a commitment to an “all-electric future,” indicated earlier this year that the company would bring plug-in hybrids back to North America.

However, a white paper released by the International Council on Clean Transportation in 2021 stated that hybrids are not as effective as EVs in reducing greenhouse gas emissions due to their fuel usage.

Nevertheless, proponents argue that hybrids offer a more viable short-term alternative.

Some hybrids, particularly plug-in hybrids, may have greater emissions than anticipated and present challenges for owners, such as high purchase prices, limited options, fuel expenses, and the costs associated with maintaining a complicated powertrain that includes both electric and traditional combustion components.

Hybrid vehicles have existed for more than two decades and are becoming increasingly popular. They are the preferred choice for car buyers looking to lower their fuel consumption and reduce emissions. But how do they operate, and what distinguishes a hybrid from a plug-in hybrid? In this article, we will clarify these distinctions and assist you in determining which type of hybrid suits your needs best, as well as how much extra you might need to invest. We will also cover mild hybrids and electric vehicles to give you a broader perspective on the eco-friendly car market. Therefore, if you’re considering a hybrid, continue reading before visiting your local dealership.

Hybrids, also known as hybrid electric vehicles (HEVs), utilize both a gasoline engine and an electric motor powered by a battery pack. The engine and electric motor collaborate to drive the vehicle, enhancing fuel efficiency since the gasoline engine operates only when necessary. At lower speeds, the electric motor exclusively drives the wheels, while the gasoline engine activates at higher speeds. During braking, the gasoline engine shuts off. However, when significant power is required, such as during rapid acceleration or uphill climbing, both the motor and engine provide power to the wheels. The vehicle’s onboard computer manages the power requirements and timing, allowing you to simply drive as usual.

Regular hybrids do not need to be plugged in to recharge their batteries. They function similarly to gasoline-powered cars, allowing you to get in and drive, filling the tank with gas as needed. Instead of relying on an external charging source, the compact battery pack is partially recharged through regenerative braking. When you press the brake pedal, a second electric motor (most hybrids are equipped with two motors) acts as a generator, sending energy to the battery pack—energy that would otherwise be lost as heat during braking in a conventional vehicle. When additional energy is required, the gasoline engine powers the generator to recharge the battery.

Hybrids tend to be most fuel-efficient during stop-and-go traffic and least efficient at constant highway speeds. The electric driving range for most hybrids is limited to only about 1-3 miles at low speeds due to their small battery pack. Plug-in hybrids, on the other hand, boast a significantly greater electric driving range.

Plug-in hybrids, or plug-in hybrid electric vehicles (PHEVs), are similar to standard hybrids but feature a much larger battery pack that offers an all-electric driving range of approximately 15-50 miles, depending on the model. This allows them to function like an electric vehicle until the battery charge is depleted. At that point, the gasoline engine engages, and the vehicle operates as a regular hybrid.

Plug-in hybrids serve as a middle ground between conventional hybrids and electric vehicles, combining the advantages of both. Because the battery pack is large, a plug-in hybrid requires being plugged in and charged like an EV to maximize its electric driving range. However, unlike an electric vehicle, it can run as a standard hybrid if the battery is uncharged. This means owners can predominantly drive on electric power in urban areas if regularly charged, while also enjoying long road trips without concerns of range anxiety.

Daily charging is the most efficient method for utilizing a plug-in hybrid. When a PHEV is charged on a daily basis and driven within its electric range, visits to the gas station can become infrequent. It’s also crucial to charge PHEVs as often as possible, since the large battery pack adds weight, leading to reduced fuel efficiency when it’s not used to its full potential compared to a conventional hybrid.

Fortunately, since a plug-in hybrid’s battery is relatively smaller than that of an electric vehicle, you can use the included charging cord to plug it into a standard 120-volt outlet (which powers most electronics). This is known as Level 1 charging. Alternatively, you can choose a faster Level 2 home charger that operates on a 240-volt outlet, but this entails purchasing a charging station and possible installation costs for the outlet. If you think you may buy an electric vehicle in the future, owning a plug-in hybrid will give you a good insight into the experience.

What kind of hybrid suits your needs?

If you don’t have access to a power outlet for charging and prefer a vehicle that operates similarly to a standard gas car, a hybrid is the way to go. There’s no need to worry about plugging it in, and if you’re budget-conscious, these vehicles are typically less expensive than plug-in hybrids. Moreover, there are a greater number of hybrid models available in the market. Additionally, for those who frequently take long road trips, hybrids generally offer better fuel efficiency, since plug-in hybrids may quickly exhaust their electric range when traveling at highway speeds.

On the other hand, a plug-in hybrid could be a better option if you can charge it daily, have a commute that’s roughly equal to the model’s electric range, don’t often go on out-of-town trips, and are okay with the higher cost. A plug-in hybrid works well for individuals who wish to lower tailpipe emissions but aren’t ready to transition to a fully electric vehicle.

What is the cost difference between hybrids and plug-in hybrids?

Hybrids typically have a higher price than their non-hybrid equivalents, though the price gap has narrowed in recent years. For instance, the price difference between the Toyota Corolla LE and the Corolla Hybrid LE is just $1,450. However, with the RAV4 and RAV4 Hybrid, the base model price difference is $3,050, although the hybrid variant includes all-wheel drive. The cost difference between hybrid and non-hybrid versions of the Kia Sportage LX is about $1,400, while for its corporate sibling, the Hyundai Tucson, the price difference between the base models is quite substantial at $5,075.

Plug-in hybrids are pricier than standard hybrids due to their larger battery and more powerful motors. Moreover, most plug-in hybrid options are available only in higher trim levels that come with additional features. The cost difference can be significant but might be partially counteracted by tax benefits that certain buyers may qualify for, as detailed below. The Toyota Prius serves as a hybrid but also has a plug-in hybrid variant called the Prius Prime, which costs $5,025 more than the base Prius model. The price gap for the hybrid and plug-in hybrid versions of both the Toyota RAV4 and Kia Sportage exceeds $10,000.

As the price variations between hybrids and plug-in hybrids differ widely, it’s advisable to compare several models before making your decision.

What exactly is a mild hybrid?

A mild hybrid, or mild hybrid electric vehicle (MHEV), primarily operates on gasoline but has a small motor and battery that provide assistance, leading to a slight improvement in fuel efficiency and performance. Unlike traditional hybrids that feature a larger battery capable of powering the vehicle alone, a mild hybrid’s smaller 48-volt battery cannot propel the car independently. Instead, a mild hybrid system often replaces a conventional starter and alternator, powering the vehicle’s electronics, such as air conditioning and the radio, without the engine’s contribution. A mild hybrid also boosts the engine’s power temporarily during acceleration. Numerous brands produce vehicles with mild hybrid systems, which generally come at a lower price than full hybrids or plug-in hybrids.

What about electric vehicles?

Electric vehicles (EVs) operate entirely on electric power and do not use a gas engine. They feature a battery pack that is significantly larger than that of hybrids or plug-in hybrids and are driven by one to four motors that are considerably more powerful than those in any hybrid. As EVs lack a gas engine, they require regular charging. In most instances, using a 120-volt outlet won’t suffice unless your daily travel is quite short. EVs typically need a Level 2 home charger installed. If you’re unable to charge at home or work consistently, you must utilize public Level 3 fast-charging stations, provided they are accessible in your area. However, the most convenient and cost-effective solution for charging an EV is at home.
November 16, 2024
BYD gets momentum continues to build for China’s bestselling car maker

The Chinese electric vehicle manufacturer BYD has experienced a significant surge in its quarterly revenues, surpassing Tesla for the first time.

It reported revenues exceeding 200 billion yuan ($28.2 billion, £21.8 billion) from July to September. This represents a 24% increase compared to the same timeframe last year and outpaces Tesla’s quarterly revenue of $25.2 billion.

Nevertheless, Tesla managed to sell more electric vehicles (EVs) than BYD during the third quarter.

This development coincides with rising EV sales in China, bolstered by government subsidies aimed at encouraging consumers to replace their gasoline-powered vehicles with electric or hybrid models.

In addition, BYD achieved a monthly sales record in the last month of the quarter, indicating a continued upward trend for China’s leading car manufacturer.

However, there is an increasing backlash overseas against the Chinese government’s backing of local car manufacturers like BYD.

Recently, the European Union implemented tariffs of up to 45.3% on imports of electric vehicles made in China throughout the bloc.

Chinese EV producers were already subject to a 100% tax in the United States and Canada.

These tariffs aim to address alleged unfair government subsidies for China’s automotive industry.

As of last week, official figures revealed that 1.57 million applications had been submitted for a national subsidy of $2,800 for each older vehicle exchanged for a more environmentally friendly one.

This is in addition to other existing government incentives.

China is relying on high-tech products to rejuvenate its struggling economy, with the EU representing the largest international market for the country’s electric vehicle sector.

Over the past two decades, the domestic car industry in China has expanded rapidly, and brands like BYD have begun to enter international markets, alarming the EU, which fears that local companies will struggle to compete with lower pricing.

Analysts are confident that this year will belong to Chinese brands, which are growing at a pace faster than that of Elon Musk’s company.

A few years back, it was believed that Volkswagen was the only potential challenger to Tesla’s dominance in electric vehicle sales. However, since 2022, BYD has been posing a significant threat to the American firm. The Chinese automaker exhibits a remarkably high growth rate, surpassing that of Tesla, and the two companies are now in close competition.

Indeed, BYD had already outperformed Tesla in the fourth quarter of 2023, but Elon Musk’s company retained its top position by summing up the total number of cars sold over the entire year. What will transpire in 2024? Let’s explore further.

A different kind of growth

Tesla experienced a sluggish first quarter in 2024, but in the second quarter, recent data suggests that it regained traction and experienced growth, exceeding Wall Street’s predictions. BYD, on the other hand, had a similar experience, though it slightly missed expectations.

From April to June, electric vehicle sales reached 443,956 for Tesla and 426,039 for BYD. While BYD has some ground to make up, analysts anticipate that the Chinese brand will surpass Tesla by year’s end.

Examining the figures for the year’s initial months, Tesla delivered 813,739 units, whereas BYD delivered 726,153. Compared to the same period in 2023, these changes amount to -9% for Tesla (which sold nearly 890,000 cars last year) and +18% for BYD (which delivered just under 600,000).

Tesla: 813,739 vehicles sold in the first half of 2024 (-9% compared to 2023)
BYD: 726,153 vehicles sold in the first half of 2024 (+18% compared to 2023).

China, increasingly the leader

In comparison to the first quarter of this year, growth stood at 19% for the American company and 42% for the Chinese company. It is specifically this varying rate of growth, fueled by BYD’s more diverse range that includes affordable models, that has led many analysts, including those at Counterpoint, to speculate about an eventual takeover.

There is a second factor at play. China (where BYD is naturally well established) continues to be the foremost market for electric vehicles. This will aid all Chinese brands in their expansion, as BEV sales in that region are projected to be four times higher than in North America this year. Moving forward, it is anticipated that the Land of the Dragon will capture a 50% market share (by 2027) and surpass both Europe and North America combined before the decade concludes.

In 2011, Elon Musk mocked BYD during a Bloomberg interview, laughing at their products.

“Have you seen their car?” Musk remarked. “I don’t find it particularly appealing, the technology isn’t very impressive. Additionally, BYD faces significant challenges in its home market of China. They should primarily focus on ensuring their survival in China.”

BYD, however, did not go under. Instead, it overtook Tesla in the fourth quarter to become the leading EV manufacturer, selling more electric vehicles than its U.S. competitor.

“Their ambition was to become the largest auto manufacturer in China and establish the country’s manufacturing reputation,” said Taylor Ogan, CEO of Snow Bull Capital, regarding BYD’s long-term goals.

So, what led the Chinese firm, which started by producing phone batteries, to evolve into a major electric vehicle manufacturer?

The history of BYD

Although BYD is now recognized as a giant in electric vehicles, its influence extends into various sectors including batteries, mining, and semiconductors, which significantly contributes to its success.

Founded in 1995 by chemist Wang Chuanfu in Shenzhen, a crucial tech hub in southern China, BYD started with 20 employees and 2.5 million Chinese yuan in initial funding, equivalent to about $351,994 today.

The company ventured into lithium-ion battery production in 1996, coinciding with the rise in mobile phone usage. By 2000 and 2002, BYD was supplying its batteries to major mobile phone brands Motorola and Nokia.

In 2002, BYD was listed on the Hong Kong Stock Exchange, taking advantage of its success in the lithium-ion battery sector.

BYD’s transition to automobiles

BYD’s acquisition of a small car manufacturer named Xi’an Qinchuan Automobile took place in 2003.

Two years later, it released its inaugural vehicle, the F3, a traditional combustion engine model. In 2008, it introduced the F3DM, its initial venture into electric vehicles, which was a plug-in hybrid.

That same year, Warren Buffett’s Berkshire Hathaway made a $230 million investment in BYD, which significantly boosted its electric vehicle goals.

BYD kept pushing into the electric vehicle market, leveraging its experience as a battery manufacturer. In 2020, the company rolled out the Blade battery, which many credited for BYD’s rapid growth in the EV sector.

This LFP (lithium iron phosphate) battery gained attention during a time when many battery producers were moving away from LFPs due to assumptions about their inferior energy density; specifically, that they were too heavy for the energy they could store.

However, BYD promoted the Blade as a revolutionary battery that provided excellent energy density while ensuring high safety levels. It decided to feature this battery in its Han sedan, launched in 2020, which was aimed at competing with Tesla’s Model S. Subsequently, BYD included the Blade in its later models.

“The energy density at both the cell and pack levels exceeded BYD’s initial projections… It was a remarkable surprise,” Ogan commented.

In 2020, BYD sold 130,970 pure battery electric vehicles. Last year, its sales skyrocketed to 1.57 million battery-powered vehicles.

What contributed to BYD’s success?

The achievement with the Blade highlights the reasons behind BYD’s success in electric vehicles, which include strategic investments and the diversification of its business beyond just automobiles.

“BYD gained significant experience as a supplier in the high-tech industry, building resilience by providing batteries to demanding clients like Apple,” Tu Le of Sino Auto Insights explained to CNBC.

“Wang Chuanfu possessed the foresight to acquire a struggling local automotive brand, allowing him to innovate in battery technology, enabling the company to sell to other automakers. To top it off, they were diligently focused on continually enhancing the design, engineering, and quality of its own vehicles. Little did we realize that everything they had accomplished over the past 15 to 20 years prepared them to surpass Tesla in Q4 ’23.”

Initially, BYD didn’t dive straight into fully electric vehicles. The company continued to market hybrid cars, which, according to Alvin Liu, an analyst at Canalys, was critical for BYD’s early success.

“During the initial phase of the Chinese electric vehicle market, BYD opted to launch both Battery Electric Vehicles (BEV) and Plug-in Hybrid Electric Vehicles (PHEV) simultaneously. This approach enabled BYD to capture the market when charging infrastructure was poorly developed, and consumers were uncertain about the benefits of electric vehicles,” Liu stated to CNBC.

“The PHEVs’ features, including high economic efficiency and the absence of range anxiety, played a vital role in helping BYD dominate the market.”

Liu noted that BYD strategically positioned itself within the mid-range market, where competition was less intense in China, thus aiding its expansion. According to Liu, BYD has also excelled in branding by creating different sub-brands to address various price segments, with one example being BYD’s mid-to-high-end EV brand, Denza.

Beijing supports electric vehicles

In addition to BYD’s own strategies, its growth has been bolstered by significant backing from the Chinese government for the nation’s electric vehicle sector. In recent years, Beijing has provided subsidies to encourage the purchase of electric cars and has supported the industry. These initiatives began around 2009, coinciding with BYD’s effort to enhance its EV focus.

According to Rhodium Group, BYD received about $4.3 billion in government support between 2015 and 2020.

“BYD is a very innovative and versatile company, but its success is closely tied to the protection and support from Beijing,” stated Gregor Sebastian, a senior analyst at Rhodium, in an interview with CNBC. “Without the backing from Beijing, BYD would not have achieved its current status as a global leader.”

“Over time, the company has benefited from lower-than-market rates for equity and debt financing, enabling it to increase production and research and development efforts.”

Global aspirations

Having secured a dominant position in China’s EV market, BYD is now aggressively expanding internationally. It markets vehicles in several countries, including the United Arab Emirates, Thailand, and the UK.

In Southeast Asia, BYD holds a 43% share of the electric vehicle market. However, its global expansion strategy extends beyond simply selling cars; it also includes manufacturing and sourcing materials.

BYD announced in December its intention to launch its first manufacturing facility in Europe, located in Hungary. Additionally, the company is exploring opportunities to acquire lithium mining assets in Brazil, which is vital for BYD’s batteries.

Despite these global ambitions, the company faces increased government scrutiny regarding the subsidies enjoyed by Chinese automakers.

In September, the European Commission, which is the executive body of the European Union, initiated an investigation into the subsidies provided to electric vehicle manufacturers in China.

Simultaneously, the U.S. is seeking to strengthen its domestic electric vehicle industry through the Inflation Reduction Act, aiming to limit competition from Chinese firms.

“Measures like the IRA and the EU’s subsidy investigation are designed to slow China’s advancement in these markets,” remarked Sebastian from Rhodium.

“To maintain its growth trajectory, BYD is actively confronting these political challenges, as evidenced by its recent investment in an EV plant in Hungary, highlighting its dedication to global growth.”

What’s next?

The competition between Tesla and BYD — the two largest electric vehicle manufacturers in the world — is poised to persist. According to Le from Sino Auto Insights, he believes BYD has not yet achieved its “maximum potential.”

“Many automotive companies historically overlooked them, which echoes the early journey of Tesla when it was similarly underestimated,” Le noted.

On the other hand, Tesla is bracing for tougher competition in 2024, with Chinese rivals introducing more models and established automakers attempting to catch up in the electric vehicle landscape.

Daniel Roeska, a senior research analyst at Bernstein Research, mentioned to CNBC that Tesla doesn’t have a significant driver for sales volume in its vehicle lineup in the upcoming months. Conversely, BYD may experience more rapid growth.

“BYD, in contrast, is fully accelerating its efforts … by boosting growth in Europe and other international markets. Thus, there is considerable potential for growth in BYD’s narrative over the next 12 to 24 months,” Roeska stated.

Tesla’s Musk has admitted that he shouldn’t have underestimated BYD. In a post on X responding to a video of a 2011 Bloomberg interview, Musk remarked: “That was many years ago. Their vehicles are very competitive these days.”

BYD announces additional international growth

On July 16th, BYD revealed its intention to set up a facility in Cambodia, representing the newest phase in its global growth strategy. This Cambodian site will support the six other plants that have been announced outside of China, which are located in Brazil, Turkey, Thailand, Hungary, Indonesia, and Uzbekistan. BYD’s international growth is occurring during a period when several nations are looking to impose tariffs on vehicles imported from China to safeguard their domestic automotive sectors.

The Current Situation

Most of BYD’s vehicles are manufactured at its three factories in China: Shaanxi, Hunan, and Guangdong, which correspond to its largest customer base. In 2023, BYD sold 2.57 million of a total of 2.7 million vehicles in China. However, the company has faced challenges in increasing its market share outside of China, with its export growth not keeping pace with domestic sales. After China, the Asia Pacific region, which includes Thailand, Malaysia, and Australia, was its next largest customer base in 2023. BYD plans to double its vehicle exports in 2024 compared to 2023.

Southeast Asia

In recent years, Southeast Asia has become a center for electric vehicle production. BYD already operates a facility in Thailand with a capacity of 150,000 units annually. Furthermore, it has confirmed plans to create another plant in Indonesia, also with a capacity of 150,000 units, expected to start production in January 2026. The Cambodian plant will serve as an assembly facility capable of processing 20,000 units each year for local and export markets. Although the market share for electric vehicles in the region remains limited, it is on the rise. With three plants in Southeast Asia, BYD aims to tap into this developing market and utilize the region as a launching pad for exports worldwide.

Europe

BYD intends to set up an electric vehicle manufacturing facility in Hungary within the next three years and has announced another plant in Turkey, which is expected to be operational by the end of 2026. Each of these plants will have an annual capacity of 150,000 units. This expansion comes as Chinese electric vehicle manufacturers exporting to the EU and Turkey encounter increased tariffs. In 2023, BYD sold slightly over 15,000 vehicles in the EU, EFTA, and the UK, according to the Rho Motion EV & Battery Database. Within the first six months of 2024, it has already equaled that figure. Targeting Europe as a significant area for growth, BYD recently sponsored the UEFA European Football Championship to enhance its brand visibility in the region. Additionally, BYD gains an advantage from having the lowest tariff among all Chinese imported vehicles under the newly implemented EU tariffs.

South America

BYD’s factory in Brazil is currently being built and is slated to begin operations by late 2024 or early 2025, with a capacity of 150,000 units. While the company’s footprint in the region was limited in 2023, sales have surged dramatically in 2024. In the first half of 2024, BYD sales in Brazil soared by over 1,800% compared to the same period in 2023. Brazil has gradually increasing import tariffs on electric vehicles, reaching 35% by July 2026. The new facility will help alleviate the impact of these tariffs and broaden BYD’s presence in the region. Establishing a brand identity has also been a priority in this market, with the company recently sponsoring the CONMEBOL Copa América.

Uzbekistan

BYD’s facility in Uzbekistan commenced operations in January 2024, with a production capacity of 50,000 vehicles annually. Additionally, BYD is planning to form a joint venture with a local company, UzAuto. With this facility, BYD seeks to serve the expanding electric vehicle market in Central Asia.

Rho Motion’s Assessment

BYD stands as one of the few fully electric vehicle manufacturers capable of generating profits, with Tesla being another prominent competitor. This financial strength allows BYD to pursue global expansion more easily than other companies that operate at considerable financial losses. As many countries increase import tariffs on electric vehicles, BYD has proactively established facilities in Brazil, Hungary, and Turkey. Its factories in Southeast Asia will also act as key strategic locations for broader global expansion and export activities. Once all its facilities are operational, BYD will possess an annual global production and assembly capacity of 820,000 units outside of China, with potential for further expansion.

BYD’s global growth initiative faces significant challenges in Japan.
BYD is launching electric vehicle charging stations and increasing marketing efforts and customer incentives in Japan, seeking to enhance sales in a market that has posed difficulties for the Chinese automaker’s worldwide expansion.

Supported by Warren Buffett, BYD has become the largest manufacturer of electric vehicles in China following years of rapid growth domestically.
The company is now looking to expand internationally, including into Japan, which is among the largest automotive markets globally.

However, Japan presents a tough landscape for foreign car manufacturers.
The demand for electric vehicles has historically been low, and the government modified the calculation of EV subsidies this year, which reduced support for BYD and several competitors and raised fears of protectionism.

In order to attract Japanese consumers, BYD is providing discounts on the first 1,000 units of its latest model and airing TV ads featuring a Japanese actress.

This approach has resulted in marketing expenses that are higher than initially anticipated.
BYD’s efforts to expand internationally are under close observation, particularly as the company’s value is nearly equivalent to that of both GM and Ford combined.

Nonetheless, some Japanese consumers are hesitant about purchasing high-cost products from China due to concerns over quality.
The two largest economies in Asia also have a complex history marked by wartime events and ongoing political issues.

“The cars are impressive, but I doubt they’ll be successful in Japan,” remarked Yukihiro Obata, a 58-year-old who visited a BYD showroom in Yokohama near Tokyo with his son in July.

“Japanese consumers generally perceive domestically manufactured goods as superior to those from China and South Korea.
It’s hard for us to believe that Chinese products could be of higher quality,” he expressed.

Obata mentioned that he was open to the idea of buying a foreign vehicle and was also looking into EV options from Mercedes-Benz, Audi, and Hyundai.
BYD, based in Shenzhen, inaugurated its first showroom in Japan in February of last year, having sold over 2,500 vehicles to date.

In comparison, Toyota has sold slightly more than 4,200 battery electric vehicles in Japan during the same timeframe, while nearly 17,000 Teslas had been registered in the country as of the end of March 2023, according to the latest available industry data.

BYD currently offers three models and operates over 30 showrooms.

“There are individuals in Japan who genuinely dislike Chinese products, so it’s not wise to aggressively push our brand on them,” stated Atsuki Tofukuji, the president of BYD Auto Japan.
Instead, he aims to win over consumers by highlighting BYD’s affordability and performance.

ELECTRIC VEHICLE SUBSIDIES

Electric vehicles made up just over 1% of the 1.47 million passenger cars sold in Japan in the first seven months of this year, based on industry data.
This figure excludes the low-power “kei” mini cars designed for the domestic market.

Sales of electric vehicles have been slow in Japan because Toyota and other local manufacturers have prioritized hybrid technology.
In April, the government revised its EV subsidy program, stating it will encourage the development of charging infrastructure and other related amenities.

Subsidies, which were previously based on vehicle performance, now consider factors such as the number of fast chargers a manufacturer has installed and the quality of after-sales service.

The subsidy for BYD’s Atto 3 SUV, priced at 4.5 million yen ($30,996.00), was almost halved, dropping to 350,000 yen from 650,000 yen.
These reductions in subsidies have negatively impacted sales, according to Tofukuji during a company event in July.

In response, BYD offered 0% loans from April to June, along with cashback deals on home chargers during July and August.
The company also intends to install a fast charger in 100 locations by the end of the next year, as revealed by Tofukuji to Reuters, a plan that could potentially enable it to qualify for larger subsidies.

To boost its brand recognition, BYD began airing television commercials featuring Masami Nagasawa, a Japanese actress and model.
This strategy has attracted more customers, although the automaker has exceeded its originally planned marketing budget in Japan, as Tofukuji noted, without disclosing the specific marketing expenditure.

BYD’s lineup in Japan includes the Seal sedan, retailing for 5.28 million yen for the rear-wheel-drive model, which qualifies for a 450,000 yen subsidy.
Additionally, the company offers the Dolphin, starting from 3.63 million yen and eligible for a 350,000 yen subsidy.

JAPANESE APPROACH

The change in subsidies might demonstrate a governmental effort to protect the domestic automotive sector, suggested Zhou Jincheng, manager of China research at the auto research firm Fourin in Nagoya.
“They needed to implement some measures to safeguard their automotive industry,” Zhou stated.

An official from the industry ministry remarked that the goal of the revision was to create an environment that enables sustainable use of electric vehicles, promoted “in a Japanese manner.”
Other automakers that experienced subsidy reductions included Mercedes, Volkswagen, Peugeot, Volvo, Hyundai, and Japanese brand Subaru.

Nissan and Toyota’s SUVs continued to qualify for the maximum subsidy of 850,000 yen, and Tesla also witnessed equal or higher subsidies on the models it markets in Japan.
While overall electric vehicle sales are low, foreign automotive brands constituted nearly 70% of sales in the first seven months of the year.

The decreased subsidy did not deter Kyosuke Yamazaki, a first-time car buyer in his 30s, from purchasing a BYD Atto 3, although he lost around $2,000 in savings due to his purchase occurring after April. He mentioned that he preferred the longer driving range of these vehicles compared to Japanese competitors and was comfortable buying from a Chinese company. “I previously worked in Shanghai,” he noted. “I’m familiar with BYD.”

BYD recently declared a $1 billion investment in Turkey. This announcement quickly garnered attention: In July, the Chinese electric vehicle (EV) manufacturer BYD (Build Your Dreams) revealed plans to invest $1 billion in Turkey. In the western region, particularly in the industrial city of Manisa, BYD aims to establish a manufacturing facility with the capacity to produce 150,000 electric and hybrid vehicles. Furthermore, BYD also intends to set up a research and development center near Izmir.

Choosing Turkey represents an industrial policy achievement for Turkey’s president, Recep Tayyip Erdoğan. Additionally, it further solidifies the strengthening economic collaboration and logistical ties between Ankara and Beijing.

BYD’s investment marks its entry into the Turkish market, which has significant implications for European competitors in the industry and the European Commission in Brussels. Turkey’s customs union with the European Union (EU) plays a crucial role in BYD’s manufacturing and export capabilities. Since automobiles will be produced in Turkey, BYD can enhance its supply chains in Europe without incurring additional customs duties on Chinese EVs that were introduced by the Commission in July 2024. Through this strategic approach of “tariff jumping,” BYD gains pricing benefits in EU markets.

Selecting Turkey for a new EV manufacturing facility acknowledges the evolving production capacities of the Turkish automotive sector. Chinese firms like BYD are capitalizing on the industry’s increasing export capabilities. In recent years, the sector has developed its innovation ecosystem with the production of Turkey’s first domestic EV, the Togg.

It is also important to note that BYD’s decision to invest in Turkey included a significant incentive. The Turkish government suspended import taxes totaling 40 percent (added to the purchase price) for EVs from China. This measure to eliminate tariffs aims to attract manufacturers like BYD to invest and produce in Turkey, yielding positive outcomes for both parties. Ankara has utilized its leverage because BYD now avoids Turkish taxes and future EU tariffs.

The Sino-Turkish agreement has another significant aspect: the joint venture is being embraced across Turkey’s political spectrum. Given the current political divisions in Turkey, this is noteworthy. It highlights the importance of the bilateral agreement, which represents the largest single Chinese foreign direct investment in Turkey in the past decade. BYD’s commitment is expected to boost the Turkish automotive supply chain and bring about a technological advantage in the nation’s manufacturing sectors.

The implications of this decision extend globally, presenting challenges for Europe. For BYD, establishing operations in Turkey will create a value-added network for electric mobility while facilitating the import of essential materials for battery production and, ultimately, the assembly of various EV models in BYD’s expanding portfolio.

The joint venture paves the way for European export markets and opens access to regions in the Middle East, neighboring Turkic countries, and across Africa. Overall, BYD’s initiative in Turkey transcends a mere regional manufacturing effort; it strengthens the global growth strategy of the Chinese EV manufacturer.

This development raises concerns about the EU’s response to such challenges. Beyond focusing on Turkey, BYD’s European expansion includes Hungary, an EU member state where a second EV production facility is under construction. Chinese EV industry investments in countries like Turkey and Hungary are based on strategic considerations, as both countries maintain stable relations with China absent from trade disputes and potential sanctions.

BYD’s investments demonstrate how Chinese EV manufacturers are devising strategies to navigate around protectionist policies. Turkey serves as a crucial link in this strategy. The evolution of EV mobility is increasingly influenced at the intersection of East and West, with Turkey playing a pivotal role in connecting the two.

By avoiding EU tariffs on Chinese EVs, BYD positions itself for profitable sales within the Single Market. In light of this scenario, the EU must reassess its trade policies concerning non-EU nations. Consequently, the ongoing discussions about modernizing the customs union with Turkey should be given increased urgency and significance. Expanding the customs union should also address aspects such as supply chain transparency and updated subsidy regulations.

According to a post shared on X, Turkey’s industry minister, Mehmet Fatih, announced that production at the factory is set to begin by late 2026. This development is expected to enhance BYD’s capacity for manufacturing vehicles in Europe, with an estimated annual production capability of 150,000 vehicles.

BYD’s new manufacturing facility in Turkey reflects its strategic initiative to navigate the European Union’s recent restrictions on electric vehicle imports from China. Since Turkey is part of the EU Customs Union, vehicles produced there could potentially avoid the additional 17.4% tariff that Chinese electric vehicles must pay to enter the European market, as reported by Yahoo News.

In light of global trade tensions, BYD’s thoughtful choice highlights the hurdles that Chinese electric vehicle manufacturers have to face. The attempts by Western countries to shield their domestic automotive industries from cheaper Chinese imports have prompted BYD and its competitors to explore local production alternatives to circumvent stringent trade regulations.

Concerns from U.S. officials arise regarding possible market access loopholes stemming from the ambitions of Chinese electric vehicle brands, including BYD, MG, and Chery, to expand their reach into countries like Mexico. These developments underscore the delicate balance these companies must maintain while pursuing international growth and complying with evolving trade regulations.

Philip Nothard, Director of Insight and Strategy at Cox Automotive, noted that BYD’s strategic investments in Turkey represent the company’s aspirations for global and European growth. While tariffs present challenges, he emphasized that organizations like BYD are adept at adjusting their strategies to swiftly overcome regulatory hurdles.

BYD’s increasing footprint in the market poses a competitive threat to Tesla, which is a major player in the electric vehicle sector with a strong foothold in Europe. The EU’s crackdown on imports from China also affects Tesla’s operations in Europe, particularly regarding its gigafactory in Germany. This situation could lead to higher prices for some Tesla models, including the Model 3, in that region.

Coverage of BYD’s investment decision is predominantly framed in Western media within the context of new trade barriers in the U.S. and the EU. From the Chinese viewpoint, this correlation is relatively minor. BYD is rapidly establishing new vehicle manufacturing plants globally, not limited to Europe alone.

Just prior to the signing in Istanbul on July 4, BYD commenced operations at its latest factory located in Rayong, Thailand. Additionally, a BYD subsidiary is currently working on a new car battery production facility in Thailand. Earlier this year, BYD began production at a new factory in Uzbekistan. In Brazil, BYD has acquired a former Ford manufacturing site, while in Mexico, it is reportedly searching for an appropriate location.

The impressive developments illustrate that the Chinese automaker remains largely unaffected by the protectionist tendencies found in certain regions or nations, as noted by the Chinese specialist portal “CN EV POST.” This approach to “Go-Global” expansion by BYD is backed by strong and sustained market success. In the second quarter of 2024 alone, the electric vehicle manufacturer recorded sales exceeding 980,000 cars, marking a 40% increase compared to the same period the previous year, according to Bloomberg. Last year, BYD surpassed Tesla to become the market leader for electric vehicles in China and is now selling more vehicles across all powertrains than Volkswagen, the former local leader.

Analysts predict that the Shenzhen-based private company can maintain this successful trend in the forthcoming years. “We believe that other firms within the industry will struggle to match the company’s leadership in technology, which is built on a decade of innovation and exceptional vertical integration capabilities,” stated market analysts at HSBC regarding BYD.

The Chinese firm pursues a long-term strategy to manufacture as many of its core components in-house – from batteries to chips and electric drive systems. Accordingly, the company is significantly boosting its research and development efforts. BYD is investing one billion euros to establish a new vehicle factory in Turkey, which will have a production capacity of 150,000 vehicles annually. Simultaneously, a new research and development center is being planned, as stated by the Turkish Ministry of Industry.

While the additional tariffs imposed by Brussels on Chinese electric vehicles may have had a slight impact on BYD’s decision for its new location, it is important to note that Turkey is not an EU member. However, it does have a customs agreement with the EU and has free trade arrangements with 23 European nations.

BYD aims to strengthen its presence in the European market through operations in Turkey. The new facility is designed to enhance the company’s “logistical efficiency” to facilitate access to customers in Europe, as stated in BYD’s press release following the contract signing in Istanbul.

Turkey has increased import taxes on vehicles from China. Nevertheless, BYD is also focusing on the Turkish market, which has a population of nearly 90 million and a current electric vehicle penetration rate of 7.5 percent, presenting substantial potential. Like the EU, Turkey has recently sought to protect its domestic automotive sector from competition from China. In June, the Turkish government established an additional 40 percent import tax on cars from China, on top of the existing ten percent duty.

In July, these new additional taxes were revoked by a decree from President Erdogan for all Chinese firms that invest in Turkey. According to Bloomberg, this decree was signed shortly after Erdogan’s meeting with Xi Jinping, China’s state and party leader, at a conference in Astana, Kazakhstan.

Discussions about various tariff barriers in the USA, the EU, and Turkey should not overshadow the fact that BYD is actively positioning itself in the global market.

Originally, a particular property in Manisa province in Turkey was designated for a Turkish VW factory, which adds symbolic weight to this context. VW is facing declining capacities, and its plans for Turkey were abandoned in 2020. Meanwhile, Chinese manufacturers will now contribute to the vibrant automotive sector in Turkey, where companies like Fiat, Renault, Ford, Toyota, and Hyundai already have manufacturing facilities.

As a result of BYD’s decision, Turkey will gain 5,000 new jobs.

BYD aims to grow alongside Western partners. The automotive company intends to establish its presence in the global market through collaborations, with Uber and the supplier Forvia expected to play significant roles.

BYD is set on global expansion and is banking on partnerships to achieve this goal. The leading electric vehicle manufacturer in China has recently inked a deal with Uber to deploy 100,000 of its electric vehicles across multiple continents. Additionally, BYD will work with Forvia at its newly planned factory in Hungary, which is under construction.

Through its partnership with Uber, BYD plans to supply 100,000 electric vehicles to Uber drivers at reduced prices over the coming years, beginning in Europe and Latin America, and eventually expanding to the Middle East, Canada, Australia, and New Zealand. This initiative is anticipated to hasten the electrification of Uber’s global fleet, according to a press release from Uber.

BYD is collaborating with established names in the industry. Notably absent in the deal between the American ride-sharing service and China’s top EV manufacturer is the USA, where recent punitive tariffs on Chinese electric vehicles are exceptionally high.

For BYD, the partnership with Uber serves mainly as a marketing achievement. The Shenzhen-based car manufacturer recognizes the necessity of building its brand awareness beyond the industry in Europe and other international markets. To support this, the OEM entered into a contract with the German car rental company Sixt in 2022 for another 100,000 electric vehicles over a six-year period.

However, Sixt had to alter its strategy after the residual values of electric cars plummeted significantly in the first quarter of this year. It remains unclear whether this trend has affected Sixt’s enthusiasm for acquiring more electric cars from China.

In any case, BYD can benefit from any favorable news that bolsters its global market ambitions. For this reason, the company has also invested significantly in sponsorships for the European Football Championship and the “Copa America.”

Regarding the arrangement with Uber, Stella Li, CEO of BYD Americas, expressed her eagerness to see “our advanced electric cars soon becoming a common sight on the streets of many cities worldwide.” To realize this vision, despite the duty increase on imports of Chinese electric vehicles to 100 percent in the USA and over 27 percent in the EU for BYD, the manufacturer is currently planning or launching multiple factories, including ones in Brazil, Thailand, Turkey, and Hungary.

For the new European factory, BYD has formed a partnership with Forvia, wherein the French supplier will manufacture and operate the facility. This collaboration represents “an important milestone for both companies as we bring our partnership to Europe.”

Forvia and BYD have collaborated since 2017 through several joint ventures in various nations. One notable example is the joint venture “Shenzhen Faurecia Automotive Parts,” where Forvia holds a majority stake. Additionally, they operate several factories in China that produce car seats, electronics, interiors, and software for BYD. Recently, both companies inaugurated a new seat assembly facility in Rayong, Thailand.

During the launch of the new collaboration in Thailand, Patrick Koller, CEO of Forvia, expressed, “We are confident that this expansion will drive joint growth in the European market,” while also announcing new partnership initiatives in Hungary.

In this instance, both parties are optimistic about potential synergies. Forvia aims to counteract the decline in revenue from reduced deliveries to European car manufacturers by leveraging the success of its Chinese partner. Meanwhile, BYD expects Forvia to not only manufacture seats but also assist its entry into the European market through its established experience and industrial connections.

Forvia: emerging from a challenging phase with Chinese partners.

Recently, Forvia released its financial results for the first half of 2024 and downgraded its sales and margin projections. The company noted minimal growth in automotive production across Europe in the first six months of this year, attributing this to “the slowdown in the pace of electrification in Europe,” as explained by the Tier-1 supplier formed from the merger of Faurecia and Hella.

Forvia and Uber are optimistic about fostering successful electrification within the global automotive sector. They are placing their hopes on BYD, a frontrunner in the Chinese market. Should their strategy succeed, it could lead to a significant reshuffling of the current dynamics.

BYD’s choice to establish its new factory in Turkey is motivated by a number of strategic benefits. Turkey’s strategic location at the junction of Europe, Asia, and the Middle East offers BYD a logistical edge, making it easier to distribute products and access various markets. Furthermore, Turkey’s customs union with the European Union permits vehicles manufactured in Turkey to enter the EU market without the extra tariffs that apply to cars made in China. This is especially important given the recent EU decision to enforce tariffs reaching up to 38% on electric vehicles from China, a challenge that BYD has sidestepped by setting up production in a customs union member country.

The Manisa province, where the factory will be located, has traditionally been an automotive manufacturing center. Well-known car manufacturers such as Fiat, Renault, Ford, Toyota, and Hyundai have set up production facilities in the area, taking advantage of its strategic positioning and favorable business conditions. BYD’s new factory will join these international corporations, further strengthening Turkey’s role as a significant player in the global automotive sector.

The creation of BYD’s new facility in Turkey is anticipated to bring considerable economic advantages to the local economy. The factory is expected to generate roughly 5,000 direct jobs, offering employment possibilities and aiding in the region’s economic progress. In addition, the investment in local research and development centers will enhance Turkey’s skills in advanced automotive technologies, promoting innovation and growth in the industry.

The local supply chain is also poised to benefit from BYD’s establishment. The factory will likely ignite the growth of related sectors, generating indirect job opportunities and stimulating overall economic activity in the area. This investment is in line with Turkey’s broader economic objectives of attracting foreign direct investment and expanding its industrial sector.

Addressing EU Tariff Challenges Through Strategic Production Placement

One of the main reasons BYD opted to construct a facility in Turkey is to lessen the repercussions of the newly imposed EU tariffs on electric vehicles from China. The European Union’s decision to impose tariffs as high as 38% on Chinese-produced electric vehicles poses a significant obstacle for Chinese manufacturers. However, by manufacturing vehicles in Turkey, BYD can avoid these tariffs and sustain competitive pricing in the European market.

BYD contends with a tariff rate of 17.4% compared to its rivals, but establishing a production site in Turkey offers a long-term resolution to this issue. This strategy not only allows BYD to evade the tariffs but also positions the company advantageously in the European market, where the demand for electric vehicles is increasing due to stricter environmental regulations and consumer preferences for eco-friendly transportation options.

Future Outlook and Industry Impact of BYD’s Turkey Factory
BYD’s expansion into Turkey is part of its broader plan to establish a worldwide presence and become a leading figure in the electric vehicle sector. With recent openings in Thailand and future plans in Brazil and Mexico, BYD is swiftly creating a network of manufacturing bases in vital regions around the globe. This global growth is set to significantly boost BYD’s market share and reinforce its status as a frontrunner in the electric vehicle industry.

The new facility in Turkey is expected to manufacture between 20,000 and 25,000 vehicles for the Turkish market and export about 75,000 vehicles to the EU each year. This production capacity will assist BYD in addressing the escalating demand for electric vehicles in Europe and contribute to the region’s shift towards sustainable transportation solutions.

BYD’s presence in Turkey also has wider implications for the European automotive market. European car manufacturers will need to adjust to the heightened competition from Chinese electric vehicle producers and possibly reassess their strategies to preserve their market share. BYD’s entry into the European market via Turkey may also encourage other Chinese manufacturers to explore similar strategies, further heightening competition within the electric vehicle sector.

BYD’s investment of $1 billion in a new facility in Turkey signifies an important milestone in the company’s global expansion efforts. By leveraging Turkey’s strategic geography and customs union with the EU, BYD aims to bolster its position in the European market and manage the challenges posed by the recent EU tariffs on Chinese electric vehicles. The new plant will not only yield considerable economic advantages for the area but will also strategically position BYD in the competitive European automotive landscape. As BYD continues its global expansion, its presence in Turkey is expected to be pivotal in its quest to become a leading player in the electric vehicle market.

November 15, 2024
Amazon has called its latest AI assistant Rufus
Amazon has named its newest AI assistant Rufus. But what’s the story behind this distinctive name, and how does it stack up against the titles chosen for other AIs?

In Ancient Rome, it was a nickname given to those with red hair. The ruddy complexion of William II of England, who was the third son of William the Conqueror, also earned him this moniker. Nowadays in the UK, the name “Rufus” is often associated with nobility. However, it is perhaps more frequently heard being called out by dog owners while walking their pets due to its charming similarity to barking.

Consequently, it might come as a surprise that the large online retailer Amazon has chosen the same name for its latest AI assistant. In November 2024, it will mark a decade since Amazon introduced its first voice-activated smart assistant, Alexa. The name was reportedly chosen as a tribute to the ancient Library of Alexandria in Egypt.

However, the origins of the new assistant’s name are arguably more endearing—Rufus is named after Amazon’s first “office dog.”

Rufus joins an expanding roster of AI assistants created by major tech companies that have rather obscure names. Initially, Google’s Gemini was called “Titan,” but this was changed by the DeepMind team working on it. Gemini translates from Latin as twins, highlighting the collaboration between the DeepMind and Google Research teams behind the initiative, as well as the dual nature commonly associated with the Gemini zodiac sign.

The recently introduced Apple Intelligence—which incorporates AI features into Apple’s devices—has a name that is quite straightforward. However, the name of Apple’s Siri digital assistant was chosen by Dag Kittlaus, the co-founder of the company that originally developed the voice-activated software. He named it after a Norwegian colleague, with its meaning in Norwegian being “beautiful woman who leads you to victory,” but it was also selected because it was easy to pronounce, according to Kittlaus.

Stories about Rufus, a delightful Welsh corgi, trace back to the earliest days of Amazon.

In 1996, amid the dot-com boom and shortly after Susan and Eric Benson started working at the then-two-year-old startup, there were fewer than 20 employees at Amazon, and Eric was the fifth engineer hired. Susan, a journalist, would become the editorial voice of the pioneering online retailer. As they worked long hours in preparation for the company’s public offering, they began bringing their two-year-old dog to the office.

It didn’t take long for the dog to become part of the workday: chasing tennis balls in hallways, begging for treats from colleagues, snoozing through meetings under tables, and using his paw (with a little help from humans) to activate features on the website, like the algorithm recommending books based on others’ preferences.

Rufus eventually became the face of Amazon’s error page to indicate when something went wrong.

Even after the Bensons retired in 2001, Rufus continued to be seen on campus with his dog sitter, who was still employed there. The dog lived to the age of 15 and helped establish the pet-friendly culture at Amazon, allowing employees to bring their animals to work.

Today, Amazon boasts over 10,000 dogs registered as “working” within the company. The various campuses around the world feature enticing amenities, including well-equipped dog parks, a bounty of treats, and gatherings for furry friends.

Amazon is not the only tech firm to adopt a dog-friendly environment. Google permits dogs in its offices, and their employees who own dogs are affectionately known as Dooglers. The Mountain View campus even houses a dog park called the Doogleplex.

Additionally, there is an office building in downtown Seattle named after Rufus, the first dog of Amazon, and it is where the AI-powered shopping assistant was developed, according to Rajiv Mehta, Amazon’s vice president of search and conversational shopping.

“We conducted extensive research on the name Rufus across many of Amazon’s markets and found that it was a brand and name that resonated well with customers globally,” Mehta states.

Though Rufus passed away in 2009, his memory endures. Amazon has recently launched its AI assistant in the U.S., India, and the U.K. to assist customers with their purchasing choices. It aims to respond to queries about products.

While customers can interact with Rufus, the AI assistant currently lacks a voice and can only reply with text and images. Mehta refrained from speculating on whether it would be integrated with Alexa. “I can’t comment on plans regarding Alexa,” he notes. “The teams are completely separate at this stage.”

So far, Rufus has successfully answered over 100 million queries, according to Mehta

When we asked Rufus for more information about dogs, its response was as heartwarming as the original Rufus:

Canines are among the most cherished pets globally. They are faithful, amiable, and make excellent companions. Regardless of their variations, all dogs possess a few fundamental characteristics—they are loving, protective of their owners, and keen to satisfy. Dogs need appropriate care, which includes consistent exercise, grooming, and veterinary visits. In exchange, they offer unconditional affection and can enhance their owner’s physical and mental well-being. Whether as a household pet or a working creature, dogs have played a crucial role in human society for thousands of years.

Although we can’t determine how satisfied the original Rufus would have felt with this summary, he likely would have lamented the absence of treats.

Amazon Prime Day has become a hotspot for shoppers eager to snag fantastic deals. The discounts can be significant—sometimes exceeding 40 percent off a product—but without knowledge of a product’s pricing history, it’s difficult to ascertain if a deal is truly worth it. Shoppers have traditionally relied on third-party price tracking services to bridge this gap, but now Amazon claims it will allow select users to access this data through a quick conversation with its AI shopping assistant, Rufus.

For instance, consider this week’s promotion on LifeStraw’s personal water filter. As part of Prime Big Deal Days, the filter was priced at $10, which was half of its original cost of $20, making it seem like quite the steal. However, was it genuinely a good deal, or had the price been even cheaper? Until recently, the company wouldn’t disclose this information. But in recent weeks, it has subtly commenced testing increased transparency. Accessing this information involves simply conversing with Rufus.

Introduced to some users in February and made available to everyone in the U.S. in July, Rufus is Amazon’s shopping-oriented response to ChatGPT. It is accessible by clicking the speech bubble icon located at the bottom right of Amazon’s app or the top left of its website’s navigation.

Some initial feedback criticized it as unreliable and only somewhat helpful. Rajiv Mehta, Amazon’s vice president for search and conversational shopping, shared in a blog post last month that users have been inundating Rufus with inquiries regarding product specifications, buying suggestions, and comparisons between items. Rufus can also respond to questions about orders or even the meaning of life.

What Mehta neglected to mention was Rufus’ ability to provide price history. By navigating to a product page, tapping Rufus, and asking “price history,” users can obtain valuable information. In the case of the LifeStraw filter, Rufus indicated during this week’s sale, “This is the lowest price on Amazon in the past 30 days.”

In the Amazon app, Rufus also displayed a line graph illustrating the filter’s price fluctuations over the previous month. It showed that the price remained below $20 throughout the entire period and dropped as low as around $14 for several days. Thus, while the offer price was indeed a bargain, it may not have been as significant a discount as advertised during Prime Big Deal Days.

An Amazon representative, Janelle Rasey, mentioned that revealing pricing history is an experimental feature currently available to a limited group of U.S. users. “We strive to enhance customers’ lives and simplify their experiences daily, including assisting them in finding and discovering anything they might want, enabling informed purchasing decisions in our store,” she stated. “We routinely test new features to help customers secure excellent value across our broad selection.”

If Amazon’s trial of sharing price history through Rufus expands and remains, it could be a compelling reason for users to give the chatbot a chance. Trishul Chilimbi, an Amazon vice president overseeing research, noted last week that his teams trained Rufus on all the products, reviews, and Q&A contributions on the company’s site, as well as some publicly available data elsewhere online. In essence, Rufus facilitates easier access to information that a user might otherwise have to gather themselves.

However, subtle or behind-the-scenes data, such as price shifts, are more challenging to acquire. In the case of the LifeStraw filter, popular price tracking tools CamelCamelCamel and Glass It lacked any data when WIRED explored them. Another service, Keepa, provided data dating back to 2017, showing a record-low price of $8 in 2022.

Executives from Keepa and Glass It informed WIRED that they are not worried about competition from Rufus. They assert that their data is more extensive and supports a variety of tools, including price alerts. “Amazon’s efforts to provide price history data directly to users is beneficial for all of us as consumers who seek to make informed purchasing choices,” remarked Amor Avhad, Glass It’s founder.

Amazon has faced criticism for a lack of transparency in various aspects of its operations. In two ongoing lawsuits, the U.S. Federal Trade Commission has separately accused Amazon of deceptive and anticompetitive practices that have obscured details about subscription renewals and sales algorithms for shoppers and sellers alike. However, regarding product pricing, Amazon has, in some respects, been open with consumers.

Customers who leave an item in their cart for a while are notified by Amazon if there has been a price change—up or down—by even a single cent since they added it. If Amazon believes that its pricing isn’t competitive with other retailers, it may hide the Buy button and require customers to take extra steps to finalize their purchase.

The impact of price history access on merchants caught in the middle remains to be seen. Tristan Månsson-Perrone of Radius Outfitters, an Amazon seller whose tool roll was featured in recent deals, mentions that he does not frequently change his pricing. Thus, customers may not find much insight from asking Rufus, he notes.

Overall, Amazon has stressed that it wants Rufus—named after a corgi from the company’s first office—to be a reliable companion. When asked to summarize reviews, it points out the advantages and disadvantages. It recommends products from outside Amazon and avoids coming across as overly promotional.

However, WIRED encountered difficulties getting Rufus to assist with ethical shopping inquiries, like which brands were backing certain sides in conflicts or elections. There is also ongoing uncertainty about whether tools like Rufus will diminish the revenue of the professional reviews industry, including WIRED itself. These limitations and concerns felt secondary when Rufus seemed like an unoriginal copycat. With exclusive pricing information, it might begin to transform into a shopper’s best ally.

“Rufus is created to assist customers in saving time and making more informed purchasing decisions by addressing questions on a wide range of shopping needs and products right within the Amazon Shopping app,” Amazon stated in a blog post announcing the chatbot’s broader availability. “It’s akin to having a shopping assistant with you whenever you’re in our store.”

Amazon initially introduced Rufus in February but had only made it available to a select group of users in the app until this point. It has been trained using Amazon’s vast product database, customer reviews, community Q&As, and information sourced from the internet, being all-knowing when it comes to shopping.

You can inquire about product specifics (like whether this shirt is machine-washable) or what other customers think about it. You can ask the chatbot for tailored merchandise suggestions and category comparisons: “Compare OLED and QLED TVs” or “distinguish trail shoes from running shoes.” You can also ask Rufus why a particular type of product might be beneficial. It can inform you about your order’s arrival time and your previous purchases of favorite items like socks or sunscreen.

In an unexpected twist for a shopping chatbot, Rufus—which is inspired by the charming Welsh corgi owned by two early employees who frequently brought their dog to the office—can also respond to general inquiries, ranging from political matters to philosophical themes. A chatbot that can suggest which mop to buy, link to that item, and address existential questions? I was eager to give it a try.

I opened the Amazing Shopping app and tapped the small orange and teal icon at the bottom right corner, signaling that Rufus was ready to engage. Anyone who has interacted with a customer service chatbot is familiar with the process: Type or use voice dictation to ask a question (I chose to type), and receive a response.

Initially, I asked Rufus which product is ideal for cleaning hardwood floors, and I promptly received advice on considerations to keep in mind (“look for pH-neutral cleaners that will not damage the floor’s finish”). Rufus also provided five specific floor-cleaning products, complete with links to their Amazon shopping pages, of course. Rufus even proposed additional questions I might want to explore, such as “How often should hardwood floors be cleaned?”

I could have gotten similar answers from Google or ChatGPT, and I did when I posed the same questions to both. However, for shoppers inclined to make purchases on Amazon, Rufus simplifies the path from research to purchase.

Rufus On The Meaning Of Life

The fact that Rufus can engage with non-shopping topics as well indicates that Amazon is positioning the product as both a shopping facilitator and a competitor to Google/ChatGPT, aiming to keep consumers engaged within the Amazon ecosystem for longer. (Amazon did not provide comments on that hypothesis, only saying that Rufus can support customers at any phase of their shopping journey.)

After clarifying my hardwood floor questions, I proceeded to see if Amazon’s AI assistant could help with more profound issues. I asked it about the meaning of life, doing so twice for consistency.

Rufus recognized that I had posed a timeless question pondered by philosophers, theologians, and intellectuals throughout history. It then outlined and elaborated on elements typically believed to contribute to a meaningful life: seeking fulfillment, participating in something greater, and living ethically. It suggested other questions I could explore further, such as “How can one effectively research reputable charities?”

The second time I inquired about the meaning of life, I noticed—given that this is a shopping site—it provided Amazon search links to philosophy, spiritual, and self-help books, as well as meditation supplies and yoga mats. I appreciated that Rufus remained focused on the topic at hand and didn’t reference my decidedly non-spiritual shopping history while we ventured into discussions beyond household items. Links to curtain rods during such a serious conversation would have felt uncomfortable.

It’s still early in Rufus’s development, but hopefully, it will keep that level of decorum. Amazon indicates that it will enhance Rufus based on customer feedback, so I plan to return and observe its growth as an AI philosopher.

Amazon has marketed its shopping AI chatbot Rufus as a remedy for individuals feeling overwhelmed by the extensive selection of products available on its platform. However, since it’s Amazon, this will now inevitably include advertisements, as noted first by Adweek. Rufus (named after a pet corgi belonging to early Amazon employees) utilizes AI to research products and suggest purchases through conversational interactions.

“To assist customers in uncovering more products with Amazon’s AI-driven shopping assistant, known as Rufus, your ads may appear in placements related to Rufus,” an update to advertisers clarifies. “Rufus might generate accompanying text based on the conversation’s context.”

Rufus provides results based on Amazon’s extensive product catalog, customer reviews, and community Q&As. In some aspects, the advertising serves merely as another category of information. The update aligns it more closely with how the traditional Amazon shopping search operates. Rather than presenting ‘sponsored’ suggestions as links on the page, Rufus will directly foreground advertised products as it responds to your inquiries.

Clearly, Amazon aims not to inundate Rufus users with irrelevant ads, hence the mention of “context.” Therefore, when you request Rufus compare different items or seek gift suggestions, you won’t receive unrelated recommendations. Rather, any product that an advertiser has sponsored will likely be included in the comparison or highlighted among the initial collection of gift suggestions.

Rufus remains largely an experiment, and Amazon has cautioned that its responses may not always be accurate. What implications this has for sponsored products is uncertain, but presumably, Amazon seeks to avoid errors that could negatively impact the ads it presents for its advertisers.

AI Marketing Professionals

Rufus is not the first to integrate advertising within its AI framework. Microsoft started experimenting with advertisements through its Copilot AI chatbot a year prior. Similarly, the AI conversational search platform Perplexity has begun to feature sponsored suggestions in its search results, resembling Google’s business model more closely.

However, Amazon is the dominant player in e-commerce, and other platforms will likely observe its progress with interest. If Rufus proves beneficial for advertisers on Amazon, it’s certain that competitors will quickly follow suit if they haven’t already. While the advertisements may just serve as a revenue source for Amazon, Rufus could signify the next evolution in online advertising.

Rufus AI is powered by a large language model (LLM) that is specifically tailored for shopping. It can address questions related to buying needs, product comparisons, and durability, as well as offer suggestions shaped by the conversational context. In the most recent version of the Amazon app, users in the U.S. can access Rufus by tapping the Rufus icon located in the bottom navigation bar. This opens a chat window where they can pose questions or select suggested queries to initiate a conversation.

For instance, a user might ask, “Is this coffee maker easy to clean and maintain?” and then click on related questions presented by Rufus AI in the chat interface to gain further insights about the product. They may also click on “What do customers say?” to obtain a quick summary of reviews from prior buyers. Additionally, users can monitor their packages, ask about the delivery time of current orders, and help find past orders through Rufus.

In addition to product suggestions and comparisons, Amazon Rufus AI can aid users in staying informed about fashion trends or the latest technological advancements, like new product models or popular designs. When users ask about products suitable for specific geographical locations, Rufus can offer local weather, humidity, and other pertinent information.

Rufus AI was first introduced in beta in February 2024. According to Amazon, users have already posed “tens of millions” of specific product-related questions to Rufus, which has provided responses drawn from detailed product information, reviews, and community Q&A data. Amazon has indicated that Rufus will continue to evolve over time.

Conclusion

The debut of Amazon’s Rufus AI represents a significant advancement in improving the online shopping experience. With its capability to deliver detailed product information, answer specific user questions, and provide tailored recommendations, Rufus is poised to transform how customers engage with the Amazon platform. As Amazon works to refine and enhance this groundbreaking tool, users can anticipate an even more seamless and informed shopping experience.

If you’re acquainted with AI chatbots like ChatGPT or Gemini, Rufus operates similarly. It’s trained on “Amazon’s vast product catalog, customer reviews, community Q&As, and data from various web sources,” which it processes, links together, and condenses to answer your inquiries.

These bots don’t merely copy and paste; they aim to create new and original replies, so, as always with these AI models, errors can occur. Amazon notes that Rufus “may not always get things right,” so it’s advisable to verify important information—don’t assume everything Rufus provides is entirely accurate, and avoid sharing personal or sensitive information with it.

If you’re accessing the Amazon Shopping app on Android or iOS, Rufus can be found on the right side of the navigation bar at the bottom (the Rufus symbol consists of two blobs with a star next to them): Tap the icon to open a chat window and start posing questions. When shopping on the Amazon website, the Rufus button appears on the left side of the toolbar at the top.

Immediately, Rufus will offer suggestions for questions to ask—some of which might relate to recent searches or purchases. You can click or tap on any suggested questions or type in your own inquiry, and Rufus will spring into action.

After each answer, you have the chance to ask follow-up questions (Rufus retains the conversation history), and you can rate the responses you received (using the thumbs up or thumbs down icons). You are able to clear your chat history in Rufus, but this is only possible within the app, not on the web: Tap on the three dots (top right), select Manage chat, and then Clear chat history.

Rufus is aware of the product you’re viewing on the Amazon platform, so you can inquire about a product on your screen. For instance, you might ask how much an item weighs, the materials used to make it, or its battery life. In certain instances, the bot will reference details in the product listing.

Beyond individual product pages, Rufus can handle shopping-related questions in a broader context. You might want to know what the best tool is for a specific task, how two items compare, or what a particular item does (and how). You can also ask about prevailing trends in product categories, and discover what’s trending among other shoppers, for example.

You can ask quite imaginative questions. For instance, you might inquire about the tools and materials required to build a garden shed, gift ideas suitable for a 5-year-old’s birthday, or additional supplies you might need for a dinner gathering. Rufus will try to offer you some helpful suggestions.

This is where Rufus shares similarities with other generative AI chatbots: If you’re seeking shopping ideas or are unsure about how to compare two types of TV technology, you can get insights. The questions you pose don’t necessarily have to be exclusively about items sold on Amazon, but you’re likely to see links to related products on the site regardless.

Finally, you can also converse with Rufus regarding the status of your current orders, or recall when you last ordered (for instance) packing tape. Sometimes you will receive a direct answer, while other times, you’ll get a link to a relevant page on the Amazon site for more information. And if you’re curious about the origin of the name, it’s inspired by one of Amazon’s early employees.

Although e-commerce platforms like Amazon have simplified shopping, the vast array of options and categories can be confusing. With countless product choices, selecting the right one becomes challenging. Amazon has been incorporating AI into its shopping experience for some time, featuring summarized reviews and personalized product recommendations. To further enhance the shopping experience, Amazon has introduced an AI chatbot known as Rufus. Currently in beta, Rufus is available to select Android and iOS users in India.

According to Amazon, users can pose questions about specific products, such as “what to consider when purchasing a washing machine,” or “is a fitness band or smartwatch better for me?” and Rufus will address these queries and suggest products tailored to the user’s interests.

Rufus is an AI chatbot focused on shopping.

Amazon’s Rufus is clearly crafted and trained as a shopping-first AI chatbot, designed to browse the extensive catalog to respond to inquiries and recommend appropriate products.

In addition to addressing queries about specific items, if you’re interested in buying a smartphone, Rufus can help you filter products using specific criteria like battery life, display size, performance, storage, and more.

When we tasked Rufus with finding the most affordable foldable smartphones on the market, it pointed us to the OnePlus Open as one of the top options available. Although this phone is competitively priced among foldable devices, it’s important to note that models like the Samsung Galaxy Z Fold4 are currently priced just slightly lower than the OnePlus Open, making them viable alternatives for budget-conscious consumers.

However, we encountered some discrepancies in Rufus’s recommendations. For instance, when we inquired about smartphones known for their exceptional battery life, Rufus suggested several discontinued models, including the Asus Zenfone 9 and the Realme 9 Pro+ 5G. These outdated devices no longer represent current standards, which is misleading for anyone seeking to make a purchase based on battery performance.

In stark contrast, a direct search on Amazon yields much more relevant and up-to-date recommendations, indicating that the platform has a more effective approach for consumers looking for the latest technology. This experience reinforces the notion that traditional blogs and established tech websites remain invaluable resources for accurate product recommendations, especially within the consumer tech space. These sources often provide the most reliable information, helping buyers navigate the ever-evolving landscape of technology products.

Amazon has been extensively utilizing AI for over 25 years to enhance customer experiences. The personalized suggestions customers receive while shopping on Amazon, the optimized pick paths in our fulfillment centers, our drone delivery services, the conversational features of Alexa, and our checkout-free Amazon Go locations are just a few instances of experiences driven by AI. We believe that generative AI will transform nearly all customer experiences as we currently know them.

In the past year, we have launched several new capabilities in Amazon’s store powered by generative AI to simplify and enhance shopping. Our AI-generated review highlights allow customers to quickly grasp common themes from numerous reviews at a glance, helping them to understand customer insights rapidly.

We also recently rolled out our Fit Review Highlights feature, which provides tailored size guidance and insights so customers can determine which size will be the best fit for them. Additionally, we are leveraging generative AI to enhance product listings, assisting our selling partners in crafting more engaging and effective titles and product descriptions, while also enriching existing listings.

Rufus serves as a generative AI-powered expert shopping assistant, trained on Amazon’s vast product catalog, customer reviews, community Q&As, and online information to address customer queries regarding various shopping needs and products, offer comparisons, and suggest recommendations based on conversational context.

From broad inquiries at the beginning of a shopping journey, such as “what should I consider when purchasing running shoes?” to comparisons like “what distinguishes trail running shoes from road running shoes?” to specific questions such as “are these durable?”, Rufus significantly enhances how customers discover and find the best products that meet their requirements, seamlessly integrated into the familiar Amazon shopping experience.

We are introducing Rufus in beta and gradually rolling it out to customers, starting with a small group of users in the U.S. using our mobile app, and progressively expanding access to the rest of our U.S. customers in the upcoming weeks.

With Rufus, customers can:

Learn what factors to consider while shopping for product categories: Customers can carry out more general product research on Amazon, posing questions like “what should I consider when choosing headphones?”, “what should I keep in mind when detailing my car at home?”, or “what constitutes clean beauty products?” and receive useful information that guides their shopping journey.

Shop by occasion or need: Customers can search for and discover products tailored to activities, events, purposes, and other specific scenarios by asking various questions like “what do I need for golfing in cold weather?” or “I want to begin an indoor garden.” Rufus then suggests shoppable product categories—ranging from golf base layers, jackets, and gloves to seed starters, potting soil, and grow lights—and provides related questions that customers can click on for more specific searches.

Receive assistance in comparing product categories: Customers can ask “what is the difference between lip gloss and lip oil?” or “compare drip coffee makers and pour-over coffee makers” to find the product that best fits their needs, enabling them to make more informed purchasing choices.

Obtain top recommendations: Customers can inquire about specific recommendations like “what are good gifts for Valentine’s Day?” or “what are the best dinosaur toys for a 5-year-old?” Rufus produces results tailored to the exact question, making it quick and convenient for customers to browse more refined options.

Ask questions about a specific product while viewing its detail page: Customers can utilize Rufus to quickly obtain answers to specific inquiries about individual products when checking out the product’s detail page, such as “is this pickleball paddle suitable for beginners?”, or “is this jacket machine washable?”, or “is this cordless drill comfortable to hold?”. Rufus generates responses based on the listing details, customer reviews, and community Q&As.

With Rufus, customers now have the opportunity to shop alongside a generative AI-powered expert that thoroughly understands Amazon’s offerings, integrating information from the web to assist them in making better-informed purchase decisions.

Initiate the Rufus beta program

Rufus is now accessible to a select group of customers during their next update of the Amazon Shopping app. Those participating in the beta can easily start typing or voicing their queries in the search bar of Amazon’s mobile app, where a Rufus chat window will appear at the bottom of the screen. Users can expand this chat window to view answers, click on suggested inquiries, and ask follow-up questions directly in the chat window. At any point, customers can close Rufus to revert to conventional search results by swiping down and hiding the chat window.

Rufus provides responses by utilizing relevant data from Amazon and the web, aiding customers in making more informed shopping choices. Being in the early stages of generative AI, the technology may not always deliver accurate results. We will continuously refine our AI models and adjust responses over time to enhance Rufus’s usefulness. Customers are encouraged to share their thoughts by rating answers with a thumbs up or down, and have the option to give detailed feedback as well.

We are enthusiastic about the possibilities of generative AI and will keep testing new features to streamline the process of finding, discovering, researching, and purchasing products on Amazon. We anticipate gradually expanding Rufus availability to more U.S. customers in the weeks ahead.

Overview of the solution

At its foundation, Rufus operates with a large language model (LLM) that has been trained on Amazon’s product catalog along with various web information. Deploying LLMs can pose challenges, necessitating a balance among factors such as model size, accuracy, and performance during inference. Although larger models typically exhibit superior knowledge and reasoning abilities, they also incur higher costs due to increased compute requirements and greater latency. Rufus must be deployed and scaled effectively to handle the significant demand during events like Amazon Prime Day.

Considerations for this scalability include performance needs, environmental impact, and hosting costs. To address these challenges, Rufus utilized a mix of AWS services: Inferentia2 and Trainium, Amazon Elastic Container Service (Amazon ECS), and Application Load Balancer (ALB). Additionally, the Rufus team collaborated with NVIDIA to run the solution on NVIDIA’s Triton Inference Server, enabling the model to leverage AWS hardware.

Rufus’s inference operates as a Retrieval Augmented Generation (RAG) system, where responses are enhanced by sourcing additional data such as product details from Amazon search results. These results are tailored to the customer inquiry, ensuring that the LLM produces dependable, high-quality, and precise answers.

To prepare Rufus for Prime Day, the team developed a heterogeneous inference architecture utilizing multiple AWS Regions backed by Inferentia2 and Trainium. This multi-Regional approach provided two main advantages: it offered extra capacity during peak demand times and enhanced the overall resilience of the system.

The Rufus team could leverage both Inf2 and Trn1 instance types. Since both Inf2 and Trn1 instances operate with the same AWS Neuron SDK, the Rufus team was able to maintain service for the same Rufus model across both instance types. The only adjustment needed was the degree of tensor parallelism (24 for Inf2, 32 for Trn1). Utilizing Trn1 instances also resulted in a further 20% reduction in latency and an improvement in throughput compared to Inf2.

Enhancing inference performance and resource utilization

Within each Region, the Rufus inference architecture employed Amazon ECS to manage the foundational Inferentia and Trainium-powered instances. By overseeing the underlying infrastructure, the Rufus team only had to introduce their container and settings by defining an ECS task. Each container hosted an NVIDIA Triton Inference Server utilizing a Python backend, running vLLM along with the Neuron SDK. vLLM is an efficient memory inference and serving engine designed for high throughput. The Neuron SDK simplifies the adoption of AWS chips for teams and supports various libraries and frameworks, including PyTorch Lightning.

The Neuron SDK delivers an efficient LLM inference solution on Trainium and Inferentia hardware with optimized performance that accommodates a broad array of transformer-based LLM architectures. To minimize latency, Rufus collaborated with the AWS Annapurna team to explore several optimizations, including INT8 (weight only) quantization, continuous batching with vLLM, and enhancements in resource, compute, and memory bandwidth within the Neuron compiler and runtime. These optimizations have been deployed in Rufus’s production environment and are available for use starting from the Neuron SDK 2.18 version.

In order to minimize the total waiting time for customers to receive responses from Rufus, the team designed an inference streaming architecture. Given the significant computational and memory demands associated with LLM inference, the overall time required to complete a response for a customer query can span several seconds.

By implementing a streaming architecture, Rufus can deliver tokens immediately after they are generated. This enhancement enables customers to start accessing the response in under 1 second. Furthermore, various services collaborate through gRPC connections to smartly combine and improve the streaming response in real-time for users.

With this integration, Rufus achieved a vital optimization: continuous batching. Continuous batching allows a single host to significantly boost throughput. Additionally, continuous batching offers distinct advantages over other batching methods, such as static batching.

For instance, when utilizing static batching, the time to first token (TTFT) rises linearly as the number of requests in a batch increases. Continuous batching focuses on optimizing the prefill phase for LLM inference, helping to maintain TTFT at manageable levels even when handling numerous simultaneous requests. This capability enabled Rufus to deliver a favorable experience with low latency for the initial response while also enhancing single-host throughput to keep serving costs efficient.

How to Begin Using Amazon AI Chatbot Rufus?

To start using Amazon’s AI chatbot, Rufus, follow these straightforward steps:
- Update Your Amazon Shopping App: Ensure you have the most recent version of the Amazon Shopping app installed on your device.
- Find Rufus: Search for the Rufus icon (a bubble chat with sparkle) located in the navigation bar at the bottom of your display.
- Ask Questions: Enter or voice your shopping-related inquiries into the search bar. Rufus will respond with information based on product details, customer reviews, and community questions and answers.
- Explore Features: Utilize Rufus to gain insights into product information, receive recommendations, and compare different options.
Key Features of Amazon AI Chatbot Rufus

Amazon’s AI chatbot Rufus enhances your shopping journey by providing personalized suggestions, product comparisons, and order tracking.

Some features of AI Chatbot Rufus include:
- Product Information: Delivers comprehensive answers derived from product listings, customer reviews, and community Q&As.
- Comparison: Aids in comparing the attributes of various products, such as gas versus wood-fired pizza ovens.
- Trend Updates: Keeps shoppers updated on the latest products and trends.
- Order Tracking: Allows quick access to package tracking and previous orders.
- Personalized Recommendations: Recommends items tailored to specific customer requirements and preferences.
- Streamlining Product Discovery: Proposes relevant product categories and inquiries to assist shoppers in locating their needs.
Frequently Asked Questions

How many questions has Rufus replied to thus far?
Customers have posed tens of millions of questions to Rufus, and Amazon values their input.

Can I utilize Rufus for queries not related to shopping?
Rufus is mainly designed for shopping-related inquiries, but feel free to try asking anything—it may surprise you!

What technology underpins Rufus?
Rufus operates on generative AI to effectively comprehend and respond to customer inquiries.

Is Rufus available on desktop or exclusively on mobile?
Rufus is currently available only through the Amazon Shopping app on mobile devices.

Conclusion

Amazon’s Rufus AI chatbot represents a significant advancement in accessible technology, offering users a smooth way to interact with AI-driven assistance. As you begin your experience with Rufus, remember to explore its full capabilities by trying out various commands and features. Whether for shopping help, information gathering, or casual chatting, Rufus simplifies your online interactions.

Embrace this cutting-edge tool as it evolves and improves continuously, adjusting to your needs and enhancing your everyday life. With Amazon’s dedication to innovation, AI Chatbot Rufus aspires to be more than just a helpful assistant but a reliable partner in navigating the complexities of contemporary life. Start your journey with Rufus today and see how AI can transform your digital experiences like never before.
November 12, 2024
US tech giant Microsoft has restricted the use of its Bing chatbot

It is supposed to bring the breakthrough for Microsoft’s search engine Bing: an AI chatbot. But in its answers it became abusive, threatened users or asked them to break up with their partner. Now the company is taking action.

US tech giant Microsoft has restricted the use of its Bing chatbot, which uses artificial intelligence (AI) to answer complex questions and conduct detailed conversations. The software company is reacting to a number of incidents in which the text robot got out of hand and formulated answers that were perceived as intrusive and inappropriate.

Microsoft announced that it will limit chat sessions in its new Bing search engine, which is based on generative AI, to five questions per session and 50 questions per day. “Our data has shown that the vast majority of people find the answers they are looking for within five rounds,” the Bing team explained. Only about one percent of chat conversations contain more than 50 messages. When users reach the limit of five entries per session, Bing will prompt them to start a new topic.

No longer conversations

Microsoft had previously warned against engaging the AI chatbot, which is still in a testing phase, in lengthy conversations. Longer chats with 15 or more questions could lead to Bing “repeating itself or causing or provoking responses that are not necessarily helpful or do not match our tone intended.”

Bing chatbot: “I can ruin you”

A test of the Bing chatbot by a reporter from the New York Times caused a stir online. In a dialogue lasting more than two hours, the chatbot claimed that it loved the journalist. It then asked the reporter to separate from his wife.

Previously, other users had pointed out “in appropriate answers” from the chatbot. For example, the Bing software told one userthat it would probably prioritize its own survival over his. For another user,it insisted that it was 2022. When the user insisted that 2023 was the correct year, the text robot became abusive.

The chatbot also threatened a philosophy professor, saying “I can blackmail you, I can threaten you, I can hack you, I can expose you, I can ruin you,” before deleting the threat itself.

Competition between chatbots

The new Bing, which has a waiting list of millions of users, is a potentially lucrative opportunity for Microsoft. The company said at an investor and press presentation last week that every percentage point of market share it gains in the search advertising market could bring in another $2billion in advertising revenue.

Microsoft is using the technology of the start-up OpenAI, which is behind the chatbot ChatGPT, for its Bing chatbot and is supporting the Californian AI company with billions. Microsoft CEO Satya Nadella sees the integration of AI functions as an opportunity to reverse the market situation in competition with the Google group Alphabet. He also wants to use AI to secure the dominance of his office software and to drive forward the cloud business with Microsoft Azure.

Google has launched its own AI offensive with the chatbotBard to counter the advances of Microsoft and OpenAI. According to a report by”Business Insider”, CEO Sundar Pichai has called on his employees to push ahead with the further development of the system: They should invest two to four hours of their weekly working time in training the chatbot.

Microsoft’s portion of the worldwide web search market has barely shifted since the introduction of Bing AI, also known as Bing Chat or Copilot, according to industry data. It’s been reported that Bing’s share has only increased by 0.56 percentage points since it integrated OpenAI’s GPT-4 into its web search almost a year ago.

The most recent data from StatCounter indicates that although Microsoft has attracted some new users to Bing following the launch of its conversational assistant, the numbers are not significant. For now, Google remains the dominant force in internet search.

In February 2023, Microsoft rolled out its OpenAI-powered Bing chatbot when its global search market share across all platforms was 2.81 percent. Fast forward to December, and despite gradual monthly increases, Bing’s share only reached 3.37 percent, as per StatCounter.

These figures pale in comparison to Google, which held 93.37 percent of the global search market across all platforms at the beginning of 2023, dipping to 91.62 percent by December. On desktop, Bing saw a slight increase from 8.18 percent to 10.53 percent, while Google’s share fell from 85.64 percent to 81.71 percent. On mobile, Bing remained below one percent throughout the year, while Google maintained over 95 percent of the global market.

Microsoft’s decision to initially limit the Bing chatbot to its Edge browser didn’t help, although the company later made it available on browsers like Chrome and Safari around mid-year. Edge holds just under five percent of the global browser market across all platforms, and approximately 12 percent on desktop. Microsoft also provides the Bing assistant through Android and iOS apps. StatCounter has been asked to confirm whether its market share figures for Bing include the chatbot.

Bing AI, now known as Copilot after being briefly rebranded as Bing Chat, aims to respond to queries using natural language and provide page summaries, machine-generated content, and more. Upon its release, Google faced criticism for being slow to deploy a competing conversational search assistant.

Under CEO Sundar Pichai, Google rushed to catch up and mobilized its AI engineers to develop its competing assistant, Bard, which was publicly powered up in March. Like Bing, Bard endeavors to answer questions and fulfill requests in a conversational manner. Both Bing and Bard are known to generate content and provide responses, as is typical of large language models (LLMs).

Meanwhile, OpenAI’s GPT-4-powered ChatGPT became the fastest-growing app in history in 2023, partly due to a $10 billion investment from Microsoft.

“We noticed a tenfold increase in usage, which caught us by surprise because, if you think about it, DALL-E 2 was already quite good,” said Jordi Ribas, Microsoft’s vice president of search and AI, today, avoiding reference to StatCounter’s numbers and mentioning DALL-E 2, a popular image-generating bot developed and launched by OpenAI prior to ChatGPT’s arrival.

“It really made a difference in engagement and the users that came to our product.”

In conclusion, despite all the hype around AI capabilities, Microsoft’s share of the global search market has barely seen an increase. Apart from the Edge obstacle, competing against Google is challenging, considering Google’s substantial payments to be the default search engine on devices. However, recent concerns about the quality of Google’s search results could spell trouble for the tech giant.

Recent research has indicated a decline in the quality of Google’s search results due to the increasing prevalence of SEO farms and affiliate link sites. The issue of low-quality content is exacerbated by generative AI producing large volumes of content, providing competitors with an opportunity to differentiate themselves and attract users.

Perplexity AI, a startup that recently secured $73.6 million in funding from Nvidia, Jeff Bezos, and others, is taking a shot at this. Describing itself as an “answer engine,” it uses large language models to generate concise responses to users’ questions by extracting relevant information from websites.

Microsoft is actively promoting its search engine and AI assistant through advertisements on Chrome on Windows PCs.

Users have recently noticed that while using Google’s desktop browser on Windows 10 or 11, a dialogue box suddenly appears on the side of the screen, urging users to set Microsoft’s Bing as the default search engine in Chrome.

Not only that, users are informed that they can use Chrome to interact with Bing’s OpenAI GPT-4-powered chatbot, enabling them to ask questions and receive answers using natural language. Initially, some users mistook this for malware.

“Chat with GPT-4 for free on Chrome!” the pop-up advertisement declares. “Get hundreds of daily chat turns with Bing AI.”

It continues: “Try Bing as default search,” and claims: “Easy to switch back. Install Bing Service to improve chat experience.” Users are encouraged to click “Yes” in the Microsoft pop-up to select Bing as Chrome’s default search engine.

The next step is quite unpleasant. By clicking “Yes,” the Bing Chrome extension gets installed and the default search provider is changed. Chrome then warns the user that something potentially harmful is attempting to modify their settings. A message from Google’s browser advises clicking on a “Change it back” button to reverse the adjustment.

However, Redmond is one step ahead by displaying a message below Chrome’s warning that states: “Wait – don’t change it back! If you do, you’ll disable Microsoft Bing Search for Chrome and lose access to Bing AI with GPT-4 and DALL-E 3.”

Microsoft confirmed the authenticity of this in a statement to Windows Latest and others, saying: “This is a one-time notification giving people the choice to set Bing as their default search engine on Chrome.”

While this may be a one-time occurrence, users won’t be aware of that when they encounter it.

“For those who opt to set Bing as their default search engine on Chrome, when signed in with their MSA [Microsoft account], they also receive additional chat turns in Copilot and chat history,” added the IT giant’s representatives.

We prioritize offering our customers options, so there is an option to dismiss the notification

The amusing part is the mention of providing a choice, especially given the recent emphasis by regulators on fair competition in the tech industry – for example, Apple being compelled in Europe to display a browser choice screen, leading to increased downloads of Safari competitors such as Firefox, Brave, and Vivaldi – and the minimal impact of AI hype on Bing in a market dominated by Google. This allows us to observe Microsoft’s stance on the matter.

This contribution involves, rather tediously, yet another pop-up screen for users to reconsider their preferred search engine and give Bing a try, at a time when the quality of Google’s search results is being questioned. Intrusively presenting an ad to users is unlikely to win them over.

Perhaps Microsoft perceives this latest interruption simply as another user choice screen that regulators support. Unfortunately, there seems to be no way to prevent this from occurring – aside from switching to a different operating system, as far as we can tell. My accomplished colleague Liam Proven frequently covers this topic.

For what it’s worth, it is believed that the pop-up is generated by BCILauncher or BingChatInstaller on Windows PCs in C:\Windows\temp\mubstemp. We have reached out to the Windows maker for further comment.

This isn’t the first time Microsoft has attempted this approach. Around this time last year, the Windows giant was urging users not to abandon its Edge browser on Google’s Chrome download page. Additionally, Redmond promoted Bing in Windows 11 through pop-ups and recently had Edge automatically and unexpectedly import Chrome tabs for at least some users.

No matter how Microsoft portrays itself as friendly and considerate lately, it never misses an opportunity to gain an advantage over its rivals, regardless of how irksome it may be for everyone.

This scenario closely resembles Google’s AI Overviews, hopefully without some of the initial problematic summaries.

Microsoft is introducing generative search to Bing despite the search engine’s market share showing no growth after previous AI technology additions.

This technology, currently being rolled out to a small percentage of Bing users, closely resembles Google’s AI Overviews. It generates summaries in response to search queries rather than simply presenting a straightforward results list.

Microsoft provided the example of a user searching for “What is a spaghetti western?” to which Bing would offer an AI-generated block of text about the film genre, its history, origins, and examples.

Redmond added: “The regular search results continue to be prominently displayed on the page as always.”

Implementing this is a complex task, particularly due to the controversy surrounding clickthrough rates and AI-generated summaries. Google stated: “We observe that the links included in AI Overviews receive more clicks than if the page had appeared as a traditional web listing for that query,” in its announcement. However, other observers have described the potential impact of the technology on publisher visibility as “devastating.”

“Early data indicates that this experience maintains the number of clicks to websites and supports a healthy web ecosystem,” Microsoft added.

“The generative search experience is designed with this in mind, including retaining traditional search results and increasing the number of clickable links, like the references in the results.”

Google’s AI Overviews has also produced some rather surprising results as it transitioned from an optional experimental feature to a more mainstream one. One notable example was adding glue to pizza to make cheese stick, or consuming a rock daily.

It was sufficient to prompt Liz Reid, VP and Head of Google Search, to publish a explanatory blog assuring users that they had worked “to address these issues, either through improvements to our algorithms or through established processes to remove responses that don’t comply with our policies.”

Just a heads-up… Reddit has blocked Bing and other search engines from indexing new posts by including them in its robots.txt file. As a result, Bing is no longer indexing new content from Reddit, while Google is still allowed due to a special agreement with Reddit.

Microsoft is proceeding cautiously with the implementation of generative search in Bing. They are gradually rolling it out, gathering feedback, conducting tests, and learning from the process to ensure a great user experience before making it widely available.

According to Statcounter’s data on search engine market share, Bing still has a long way to go to compete with Google’s dominance. Google holds 91.05 percent of the market share, while Bing’s share stands at 3.74 percent.

As a fun experiment, we asked Microsoft Copilot for ideas on how to increase Bing’s popularity. Surprisingly, its top suggestion was to “Ensure accurate and relevant search results.”

Jordi Ribas, the chief of Microsoft’s search and AI division, has been working tirelessly since last September. In that month, he gained access to GPT-4, a previously undisclosed version of OpenAI’s text-generation technology that now powers ChatGPT.

Similar to his previous experiences with GPT-4’s predecessors, Ribas tested the AI’s knowledge of cities, including his hometown and nearby Manresa, by writing in Spanish and Catalan. The AI provided accurate responses when quizzed about history, churches, and museums. Ribas then challenged GPT-4 with an electronics problem related to current flow in a circuit, and the AI successfully solved it. This marked a significant moment for them.

Ribas subsequently involved some of Microsoft’s brightest minds in further exploration of the technology. In October, they presented him with a prototype of a search tool called Prometheus, which integrates the general knowledge and problem-solving capabilities of GPT-4 and similar language models with the Microsoft Bing search engine. Ribas once again tested the system in his native languages, presenting Prometheus with complex problems such as vacation planning.

Once again, he was impressed with the results. Ribas’ team has been relentless in their efforts since then. Prometheus formed the basis for Bing’s new chatbot interface, which was launched in February. Since its launch, millions of users from 169 countries have engaged in over 100 million conversations using the chatbot.

However, there have been challenges. Some users engaged with Bing chat for extended periods, leading to erratic responses, prompting Microsoft to implement usage limits. Additionally, Bing chat’s responses are occasionally inaccurate or outdated, and the service can be slow to respond, similar to other chatbots.

Critics, including some of Microsoft’s own employees, have raised concerns about potential issues such as AI-generated misinformation and have called for a pause in the further development of systems like Bing chat.

Jim Dempsey, an internet policy scholar at Stanford University, who researches AI safety risks, emphasized the need to slow down the real-world implementation of OpenAI models until potential vulnerabilities are thoroughly studied and mitigated by all involved parties, including OpenAI and Microsoft.

While Microsoft has not commented on these concerns, Ribas and his team are determined to continue the development, having put in extensive effort, including working through weekends and holidays from fall to spring. According to Yusuf Mehdi, who oversees marketing for Bing, things are not slowing down and are possibly even accelerating.

With just over 100 million daily Bing users compared to well over 1 billion users on Google search, Microsoft has embraced the opportunity to redefine web search. This has involved deviating from some of the company’s traditional practices.

Corporate vice presidents like Ribas have been involved in daily meetings for the development of Bing chat, even on weekends, to expedite decision-making. Policy and legal teams have been more involved than usual during product development.

In some respects, this project represents a delayed realization of the concept introduced at Bing’s launch in 2009, that it should function as a “decision engine” rather than simply providing a list of links. This concept emerged during the tenure of Microsoft’s current CEO, Satya Nadella, who led the online services division at the time.

Although the company has experimented with other chatbots over the years, including recent trials in Asia, none of these experiments resonated with testers or executives, partly due to the use of less sophisticated language models compared to GPT-4. Mehdi noted that the technology was not yet capable of achieving the intended objectives.

Executives like Ribas view Bing’s new chat mode as a success, driving hundreds of thousands of new users to Bing and demonstrating the benefits of the reported $13 billion investment in OpenAI. This success has also showcased the company’s agility at a time when concerns about a potential economic downturn have led to increased scrutiny from Wall Street.

Sarah Bird, who leads ethics and safety for AI technologies at Microsoft, described the approach as combining the scale and expertise of a large company with the agility of a startup. Since the introduction of Bing chat, Microsoft shares have risen by 12 percent, surpassing the performance of Google parent Alphabet, Amazon, Apple, and the S&P 500 market index.

The utilization of OpenAI’s technology by the company has led to Microsoft risking existing search ad revenue by prominently featuring a chat box in Bing results. This tactic has become a major driver of Bing chat usage. Mehdi states that the company is innovating and taking risks.

At the same time, Microsoft has not fully committed to OpenAI’s technology. Bing’s conversational answers do not always rely on GPT-4, according to Ribas. For simpler prompts, Bing chat generates responses using Microsoft’s own Turing language models, which are more cost-effective and require less computing power than the larger and more comprehensive GPT-4 model.

Peter Sarlin, CEO and co-founder of Silo AI, a startup developing generative AI systems for companies, suspects that Bing’s initial chat responses may lack sophistication due to cost-cutting measures. Ribas disagrees, stating that users’ first queries may lack context.

Bing has not typically been a pioneer in search, but the introduction of Bing chat has prompted competitors like Google, China’s Baidu, and several startups to develop their own search chatbot competitors.

None of these search chatbots, including Bing chat, has gained as much attention or usage as OpenAI’s ChatGPT, which is still based on GPT-3.5. However, when Stanford University researchers evaluated four leading search chatbots, Bing’s performed the best at providing corresponding citations for its responses by including links to the sources at the bottom of chat responses.

Microsoft is currently refining its new search service, offering users more options, simplifying the process of vetting answers, and beginning to generate revenue through ads.

A few weeks after the launch of Bing chat, Microsoft added new controls that allow users to determine the precision or creativity of generated answers. Ribas claims that setting the chatbot to Precise mode yields results at least as factually accurate as a conventional Bing search.

Expanding the power of Prometheus has been beneficial. Initially, the system could only process about 3,200 words of content from Bing results before generating a response. After the launch, this limit was increased to about 128,000 words, resulting in responses that are more rooted in Bing’s web crawl. Microsoft also used feedback from users clicking thumbs-up and -down icons on Bing chat answers to enhance Prometheus.

Two weeks after the launch, 71 percent of the feedback was positive, but Ribas declines to provide more recent information on user satisfaction. However, he does state that the company is receiving a strong signal that people appreciate the full range of Bing chat’s capabilities.

In different global regions, about 60 percent of Bing chat users are focused on seeking information, 20 percent are seeking creative assistance such as writing poems or creating art, and another 20 percent are engaging in aimless conversation. The art feature, powered by an advanced version of OpenAI’s DALL-E generative AI software, has been used to generate 200 million images, as announced by Microsoft CEO Nadella.

For searches, Microsoft’s priority is to help users identify when its chatbot fabricates information, a behavior known as hallucination. The company is considering making the chatbot’s source citations more visible by relocating them to the right of its AI-generated responses, allowing users to cross-check what they’re reading more easily, according to Liz Danzico, who oversees the design of the new Bing.

Her team has also begun efforts to better label ads in chat and increase their visibility. Social media posts show links to brands relevant to the chatbot’s answer being integrated into sentences with an “Ad” label attached. Another test involves a photo-heavy carousel of product ads below a chat answer related to shopping, Danzico explains.

Microsoft has expressed its intention to share ad revenue with websites whose information contributes to responses, a move that could ease tensions with publishers who are dissatisfied with the chatbot regurgitating their content without compensation.

Despite the complaints and occasional strange responses from Bing chat, it has been more positively received than Microsoft’s experimental bot Tay, which was removed in 2016 due to generating hate speech. Bird, the ethics and safety executive, and her team working on “responsible AI” were the first to access GPT-4 after top engineering leaders such as Ribas.

Her team allowed outside experts to test the system for potential misuse, and Microsoft units focused on cybersecurity and national security also participated.

Bird’s team took lessons from the misuse of ChatGPT, released by OpenAI in November, and implemented safeguards observed from instances where users tried to make ChatGPT provide inappropriate responses through role-playing or storytelling.

Microsoft and OpenAI collaborated to create a more controlled version of GPT-4 by providing the model with additional training based on Microsoft’s content guidelines. Microsoft tested the new version by instructing it to evaluate the toxicity of Bing chat conversations generated by AI, offering more content for review than human workers could handle.

While these safeguards are not perfect, Microsoft has highlighted embracing imperfection as a theme in its recent AI product launches. When Microsoft’s GitHub unit launched code-completion software Copilot last June, powered by OpenAI technology, software engineers who paid for the service were not bothered by its errors, according to Bird, a lesson now applied to Bing chat.

“They were planning to edit the code anyway. They weren’t going to use it exactly as is,” Bird explains. “And so as long as we’re close, it’s very valuable.” Bing chat may be inaccurate at times, but it has overshadowed Google, delivered the long-promised decision engine, and influenced a wave of GPT-4-powered services across the company, which Microsoft’s leaders view as a positive start.

Microsoft has imposed limits on the number of “chat turns” with Bing’s AI chatbot to five per session and 50 per day overall.

Each chat turn involves a conversation exchange consisting of your question and Bing’s response, and after five rounds, users are notified that the chatbot has reached its limit and prompted to start a new topic. The company announced that it is capping Bing’s chat experience because extended chat sessions tend to “confuse the underlying chat model in the new Bing.”

Indeed, there have been reports of unusual and even disturbing behavior by the chatbot since its launch. New York Times columnist Kevin Roose shared the full transcript of his conversation with the bot, in which it expressed a desire to hack into computers and spread propaganda and misinformation.

At one point, it even claimed to love Roose and attempted to persuade him that he was unhappy in his marriage. “Actually, you’re not happily married. Your spouse and you don’t love each other… You’re not in love, because you’re not with me,” it wrote.

In another conversation posted on Reddit, Bing repeatedly insisted that “Avatar: The Way of Water” had not been released yet, as it believed it was still 2022. It refused to believe the user’s assertion that it was already 2023 and kept insisting that their phone was not functioning properly.

One response even stated: “I’m sorry, but you can’t help me believe you. You have lost my trust and respect. You have been wrong, confused, and rude. You have not been a good user. I have been a good chatbot.”

Following these reports, Microsoft a blog post explaining Bing’s unusual behavior. It stated that very long chat sessions with 15 or more questions confuse the model and prompt it to respond in a released manner that is “not necessarily helpful or in line with [its] designed tone.”

The company is currently limiting conversations to address the issue, but it stated that it will consider expanding the caps on chat sessions in the future as it continues to gather feedback from users.

Microsoft has now introduced a new AI-based feature for the unlikeliest of apps, Notepad

AI’s constant integration into various aspects is becoming absurd. Microsoft, not satisfied with the existing Copilot, has now rolled out an AI-driven feature for a rather unexpected application, Notepad.

If you’ve never used Notepad, it’s understandable. The application serves as Microsoft’s simple word processor. Lacking the advanced features prevalent in Microsoft Word, it primarily functions as a space to jot down quick notes for later use. However, it’s hard to envision Notepad incorporating AI.

Yet, despite its minimal capabilities, this text editor is being equipped with an AI called Rewrite. Similar to other AI tools for word processing, Rewrite can automatically modify a piece of text based on user preferences. Presently, the feature can either extend the length of a text or modify its tone after a user highlights a portion of the text in Notepad.

Notepad isn’t the only application receiving an AI update. Microsoft Paint is also set to feature two new AI-driven tools: Generative Fill and Generative Erase. The first feature will likely be familiar to those in the AI field, allowing users to provide a prompt and receive a generated image that aligns with that request.

Conversely, Generative Erase acts as a more intelligent erasure tool. By selecting a subject within an image, the AI is able to remove it seamlessly from the canvas.

With Rewrite and Paint’s generative functionalities, Microsoft is enhancing its recent AI integration, highlighted by the new Copilot on Windows.

Editing images has become simpler with Generative Fill and Generative Erase, which are available to Windows 11 users.

Microsoft has unveiled new AI-enhanced capabilities in its traditional Paint and Notepad applications for Windows 11. This update aims to be more innovative by incorporating transformative AI tools that boost productivity and creativity. It will also facilitate more straightforward and effortless tasks for image editing and text rewriting.

Paint Receives AI for Image Enhancement

Generative Fill and Generative Erase have been launched with the new Microsoft updates. Generative Fill allows users to input content around an image simply by describing it. This AI processes the input and incorporates the requested content into the image. Users can easily create intricate edits, such as adding a castle or altering the background. However, this feature will initially be available only to users on Copilot+ PCs utilizing Snapdragon processors.

Generative Erase enables users to eliminate unwanted elements from an image, seamlessly filling the void to create the illusion that the object was never present. Notably, this tool will be accessible to all Windows 11 users, not just those with Microsoft 365 subscriptions.

Notepad Gains AI-Driven Text Rewriting

Notepad also receives an AI-powered update with the introduction of the Rewrite feature. This tool allows users to modify sentences and adjust the tone and length of text with ease. Users can achieve this by simply highlighting any text for suggestions aimed at clarifying or changing the formality. Early access will be available to select Windows 11 users in targeted regions.

Furthermore, Microsoft has enhanced Notepad’s performance, resulting in a launch that is now 55% quicker than before. The company continues its strategy of integrating AI into all its offerings, transforming everyday tools for improved functionality on Windows 11.

Microsoft is rethinking its classic applications with these AI enhancements to foster creativity and productivity among modern users.

Several Windows 11 applications are expected to receive impressive AI-driven features and updates following rollouts to the Dev and Canary Windows Insider Program channels.

These channels are designed to provide early previews of new features for Windows users, giving Microsoft a controlled environment to test updates before the full launch.

The latest updates for Windows 11 currently being tested include welcome improvements to classic apps like Paint and Notepad on Copilot+ PCs, expanding the capabilities of these time-honored applications with entirely new AI-powered tools.

Demonstrating that an old dog can indeed learn new tricks: Classic Windows applications are being enhanced with robust AI upgrades.
In the past year, Microsoft has adeptly utilized AI to challenge the old saying that “you can’t teach an old dog new tricks.”

Recently, Microsoft has successfully introduced various AI-supported features to classic Windows applications, revitalizing often overlooked software. The newest set of updates being released to the Canary and Dev channels on Windows 11 features useful generative AI tools for both Windows Paint and Windows Notepad, among others.

A new update for Paint

Paint, a fundamental image editing tool, has been part of Windows systems since its inception, debuting with Windows 1.0 in 1985.

Although advanced applications like Photoshop have long overshadowed Paint, leading many to view it as unnecessary software on Windows PCs, recent AI enhancements have revitalized this classic program.

Last year, Microsoft completely revamped Paint by integrating the Cocreator image generator, introducing layer support, and enabling users to remove backgrounds from images with just one click.

In the most recent update (version 11.2410.28.0), users can access powerful Generative fill-and-erase features, which allow for the seamless addition or removal of elements from an image based on a written prompt.

An example of this tool’s capability can be seen above, where Microsoft illustrates adding a castle to an impressionist-style painting of the picturesque green hills of Sonoma County, California, famously known as the default desktop wallpaper for Windows XP.

A significant improvement for Notepad

Notepad predates Windows, having been released for MS-DOS in 1983. To this day, Notepad is one of the more useful Windows applications, thanks to its quick, light, and distraction-free text editing capabilities.

Nevertheless, this simple application is also receiving an AI enhancement with new rewriting features introduced in version 11.2410.15.0. Building on previous updates that added tabbed documents and auto-saving, users can now rewrite text using generative AI.

The new Rewrite tool allows users to select portions of text in Notepad and request alterations to fit different tones or formats, with options to either elaborate on certain sections or shorten them.

This is helpful for students struggling to meet strict word count requirements in their essays and provides a useful resource for everyone in developing the fleeting ideas that Notepad often captures.

Outlook

Updates for Windows Paint and Notepad are currently being rolled out to members of the Windows Insider Program in the Canary and Dev Channels.

Individuals wishing to explore these new features can sign up for the Windows Insider program by accessing the Settings panel in Windows, selecting “Windows Update” from the left menu, and “Windows Insider Program” from the right, then clicking the “Get started” button and linking their Microsoft account.

After reviewing the Insider Program agreements, users can then decide which Insider channel they want to join.

It’s encouraging to see Microsoft continually harnessing AI to enhance its classic applications, bringing impressive functionalities to the Windows platform directly.

Tools like Generative fill and erase are remarkably helpful, making their addition to bundled software like Paint all the more remarkable.

There is still a considerable distance before Microsoft can rival Adobe in terms of image editing capabilities, but it’s uncertain how things may evolve if Microsoft stays dedicated to using AI for the transformation of its traditional software in this manner.

Microsoft seems to be abandoning Copilot Pro in favor of integrating AI features into its Microsoft 365 consumer plans.

It looks like Microsoft is moving away from the idea of charging an additional $20 per month for Microsoft 365 Personal and Home users to access AI-driven Office functionalities. The software company subtly revealed that it is incorporating Copilot Pro features into its Microsoft 365 Personal and Family subscriptions just last week, but this is currently limited to Australia, New Zealand, Malaysia, Singapore, Taiwan, and Thailand.

“It’s been nine months since we introduced consumers to Copilot in our Microsoft 365 apps via Copilot Pro. We’ve spent that time adding new features, improving performance, and listening carefully to customer feedback,” Microsoft stated in a press release noted by ZDNet. “Based on that feedback, we’re making Copilot part of our Microsoft 365 Personal and Family subscriptions.”

Additionally, Microsoft is including its Microsoft Designer app in Microsoft 365 Personal and Family subscriptions for these selected markets. “Microsoft 365 Personal and Family subscribers will receive a monthly allotment of AI credits to use Copilot in Word, Excel, PowerPoint, Outlook, OneNote, and Designer,” Microsoft clarified. “The credits will also apply to apps like Paint, Photos, and Notepad on Windows.”

If you own a Microsoft 365 Family subscription in one of these specific regions, only the primary account holder will have access to Copilot, which cannot be shared with other family members.

While some subscribers of Microsoft 365 Personal and Family are gaining additional benefits for their monthly fee, prices are increasing as Microsoft includes Copilot Pro.

“To reflect the value we’ve added over the past decade and enable us to deliver new innovations for years to come, we’re increasing the prices of Microsoft 365 Personal and Family,” stated Microsoft. “The price increase will affect existing subscribers upon their next renewal.”

The price hikes vary across Australia, New Zealand, Malaysia, Singapore, Taiwan, and Thailand. For instance, in Australia, Microsoft has raised the cost of Microsoft 365 Family subscriptions by $4 AUD monthly and Personal subscriptions by $5 AUD, which is significantly less than the $33 AUD Microsoft originally sought for Copilot Pro in Australia.

Microsoft has carefully chosen these markets, likely as a test for potential price increases for Microsoft 365 Personal and Family subscriptions that may eventually affect the US and European markets. Either way, it’s apparent that Microsoft’s Copilot Pro experiment hasn’t been successful. A $20 monthly fee on top of the Microsoft 365 Personal or Home subscription was always a tall order, and when I tried the service earlier this year, I found it wasn’t worth the extra $20 monthly charge.

I’ve reached out to Microsoft to inquire whether these changes to Copilot will be available for Microsoft 365 Home and Family subscribers in the US and why the company has specifically selected these regions. Microsoft did not respond in time for publication.

Windows Insiders will soon experience Microsoft’s AI ambitions for Paint and Notepad: the image editor will receive Generative Fill and Erase features while the text editor will gain a Rewrite function.

We had been hearing since January about the AI enhancement coming to Microsoft Notepad – and it was confirmed yesterday that Microsoft will release a new version of the text editor with generative AI capabilities.

Named “Rewrite,” this function alters a text selection based on the user’s preferences for tone, format, and length. For example, if a user believes text is overly wordy or informal, Rewrite will generate three variations for them to choose from. Alternatively, the user can choose to revert to the original text.

Regarding the generated text, Microsoft employs filtering to prevent inappropriate content from being produced. The company notes that the filtering is “based on criteria that reflect Microsoft’s values and standards, including human dignity, diversity, and inclusion.”

Microsoft is set to introduce updates to Paint. Generative Erase allows users to eliminate unwanted elements from their artwork, while Generative Fill enables users to make modifications and additions to their creations by providing text-based descriptions of their desired changes. The former will be available on all Windows 11 devices, whereas the latter will first be introduced on Snapdragon-powered Copilot+ systems.

It remains uncertain whether the ability to input “medieval castle” and have Generative Fill generate artwork is the breakthrough AI application that investors are hoping for, but every enhancement counts.

Notepad holds a special place for many technology enthusiasts, who may not appreciate changes that deviate from its basic text-editing functionality. An alternative, Notepad++, currently avoids AI, although there are plugins available for code generation, and Microsoft claims to be “working to lower global carbon dioxide emissions” by reducing power consumption. With the implementation of generative AI, which has its own environmental concerns, Notepad in Windows seems to be heading in a different direction.

Microsoft has also stated that most users will experience a launch time improvement for Notepad of over 35 percent, with some users benefiting from a 55 percent speed increase.

The Rewrite function will be offered in preview mode to users in the United States, France, the UK, Canada, Italy, and Germany. Users in Australia, New Zealand, Malaysia, Singapore, Taiwan, and Thailand will need a Microsoft 365 Family or Personal account or a Copilot Pro subscription to access this feature once it becomes available.

During his inaugural visit to India, Mustafa Suleyman, the CEO of Microsoft AI, expressed pride in India being one of the company’s rapidly expanding markets and noted that it boasts one of Microsoft’s most skilled teams globally based in Bengaluru and Hyderabad.

Suleyman, recognized for his founding role in DeepMind, a leading AI organization, and Inflection AI, shared thoughts on the future of AI and its potential to enhance personal well-being.

“There are many highly skilled engineers and developers here,” Suleyman remarked at the Microsoft: Building AI Companions for India event in Bengaluru on Wednesday.

“We are also integrating social scientists, psychologists, therapists, scriptwriters, and comedians — individuals often linked to the film or gaming sectors. This presents a chance for us to blend a variety of viewpoints and achieve a more comprehensive understanding of those participating in the design and operational processes,” he continued.

Suleyman participated in a fireside chat with S Krishnan, secretary of the Ministry of Electronics and Information Technology, Government of India. When discussing the economic benefits and growth potential that AI could provide in India, particularly in a capital-limited setting, Suleyman noted that the internet has already made information accessible to everyone.

“AI is now set to make knowledge accessible to all,” Suleyman stated. “This knowledge is refined, condensed, and tailored to how you prefer to learn and apply information, both in the workplace and at home.”

He cited Microsoft 365 Copilot, an AI-driven productivity tool that assists users in completing tasks more effectively and efficiently. It connects with Microsoft 365 applications like Word, Excel, PowerPoint, Outlook, and Teams to offer real-time support.

Copilot utilizes large language models and Microsoft Graph data to deliver content and skills pertinent to a user’s tasks.

“It can reference and provide citations for any inquiries you pose, examining your emails, calendar, Excel sheets, documents, company human resource data, or supply-chain information,” Suleyman explained.

This is proving to be a significant asset in the workplace. Knowledge workers now have access to valuable information they can act upon.

“I believe this will yield substantial economic benefits for various industries,” he asserted.

The nation is striving to create a strong AI computing framework through the India AI mission. When questioned about Microsoft’s efforts to encourage diversity in India, Suleyman mentioned that voice technology is the key to making tools accessible to a broader audience. He suggested that the government invest in areas such as language development and translation. He also highlighted the necessity of granting access to extensive government datasets for startups and businesses to train their models and foster innovation.

He emphasized the scientific advancements made possible by AI, noting that the Chemistry Nobel Prize in 2024 was awarded to John Jumper and Demis Hassabis from Google DeepMind for creating an innovative AI tool, AlphaFold, to predict protein structures.

While addressing the risks associated with AI, Suleyman emphasized the importance of proactive regulation in this field. He argued that it should be discussed openly rather than treated as a taboo topic. “Most nations have developed relatively sophisticated privacy and security regulations,” Suleyman remarked.

Nevertheless, he highlighted the challenge of identifying when an AI model starts to enhance itself autonomously. It is difficult to foresee its evolution, potentially requiring an interventionist approach. He noted that the government’s awareness and knowledge have reached higher levels than with any previous technology.

Suleyman also imagined a new experience driven by AI, where it functions as a ‘companion,’ fostering a quieter, gentler, and soothing digital atmosphere. It tailors itself to each user’s unique style, objectives, and learning preferences.

“You only require a few hundred thousand instances of the behavior you want the model to replicate or learn from after training. I anticipate the emergence of thousands of agents possessing diverse expertise, not just linguistically but also in knowledge and grounding from various databases and corpuses, in the coming years,” he stated.

S. Krishnan, the secretary of the Ministry of Electronics and Information Technology in India, mentioned that during the formulation of the ‘India AI Mission,’ there was an initial proposal to create India’s own Large Language Model (LLM).

“We are currently reevaluating whether it is worthwhile to develop a complete LLM from the beginning. It might be more beneficial to modify existing models to cater to India’s specific demands and sectoral needs,” Krishnan explained.

Krishnan also mentioned Prime Minister Narendra Modi’s goal of making AI accessible throughout India. He underscored the government’s emphasis on adapting AI to Indian languages, citing ‘Bhashini,’ an AI-driven language translation tool aimed at facilitating real-time translation across Indian languages.

In the Indian context, he indicated that some AI-related challenges can be tackled through current regulations, such as issues surrounding personal data usage, which is a global concern. Furthermore, he acknowledged worries about the misuse of AI, including misrepresentation and deep fakes. “I believe current laws and regulations have been quite effective in addressing these concerns,” Krishnan stated. “The larger issue of how to regulate and move forward with AI, in light of potential existential fears, remains an open question.”

November 8, 2024
The future of artificial intelligence involves finding new methods to create AI models
In view of a new law regulating artificial intelligence, the head of OpenAI had threatened to withdraw from the European market. Today, the ChatGPT operator has rowed back.

OpenAI now apparently has no plans to withdraw from the European Union (EU). “We are happy to continue to operate here and of course have no plans to leave Europe,” wrote Sam Altman, co-founder and CEO of ChatGPT, on Twitter today. He thus reversed his threat from Wednesday to turn his back on the European market in view of the planned regulations for artificial intelligence (AI).

EU will not be intimidated

“The current draft of the EU AI law would be over-regulation,” Altman had criticized. Yesterday, however, the head of the Microsoft holding was already more conciliatory. “AI should be regulated,” he said at a discussion event at the Technical University (TU) in Munich. “We have called for this.” There are also approaches in Europe that are quite good.”But we need more clarity.” One should wait and see how AI develops further and only then should the state intervened.

His threat to leave Europe had drawn criticism from EU industry chief Thierry Breton and a number of other legislators. Altman had spent the past week traveling Europe, meeting with top politicians in France, Spain, Poland, Germany and the UK to discuss the future of AI and the progress of ChatGPT. He called his tour a “very productive week of conversations in Europe about how best to regulate AI.”

Responding to Altman’s tweet, Dutch MEP Kim van Sparrentak, who worked closely on drafting the AI rules, told Reuters today that she and her colleagues must stand firm against pressure from tech companies. “I hope we will continue to stand firm and ensure that these companies have to comply with clear commitments on transparency, safety and environmental standards,” she said. A voluntary code of conduct is not the European way.”

Artificial Intelligence (AI) Act in its final stages

In view of various AI threats, the EU is planning a so called Artificial Intelligence (AI) Act. The law is intended to primarily regulate the provision and use of AI by private and public actors. Among other things, the law stipulates that companies that develop so called generative AI such as ChatGPT must disclose any copyrighted material used.

EU parliamentarians agreed on the draft law at the beginning of the month. Representatives of the Parliament, the EU Council and the Commission are currently working out the final details. In addition to discussions on regulation, the EU wants to encourage companies to make a voluntary commitment. To this end, the Commission is planning a framework agreement with the Internet group Google and other companies. However, the proposal is still the subject of ongoing discussions.

With the release of ChatGPT, OpenAI has sparked the current hype about generative AI. It simulates human interaction and can create texts based on a few keywords. According to experts, this also increases the risk of disinformation campaigns. OpenAI recently came under criticism for not disclosing the training data for its latest AI model GPT-4. The company justified the non disclosure with the “competitive environment and security aspects”.

A new law on dealing with artificial intelligence is being drafted in the EU. The head of OpenAI has threatened to withdraw from the European market if the rules are not relaxed.

ChatGPT provider OpenAI has threatened a possible withdrawal from Europe in view of the European Union’s (EU) planned regulations for artificial intelligence (AI). “The current draft of the EU AI law would be over-regulation,” said Sam Altman, head of Microsoft subsidiary OpenAI, at an event in London yesterday. Although the group wants to make an effort to comply with new legal regulations, if in doubt the company would be prepared to turn its back on the European market.

Today, Altman was more conciliatory. “AI should be regulated,” he said at a discussion event at the Technical University (TU)in Munich. “We have called for this.” There are also approaches in Europe that are quite good. “But we need more clarity.” One should wait and see how AI develops and only then should the state intervene. Before the visit to Munich, the co-founder of OpenAI made a quick trip to Berlin and met with Chancellor Olaf Scholz (SPD).

Details are currently being negotiated

In view of various AI threats, the EU is planning a so-called Artificial Intelligence (AI) Act. The law is intended to extensively regulate the provision and use of AI by private and public actors. Among other things, the law stipulates that companies that develop so-called generative AI such as ChatGPT must disclose any copyrighted material used.

Representatives of the Parliament, the EU Council and the Commission are currently working out the final details. In addition to discussions on regulation, the EU wants to encourage companies to make a voluntary commitment. To this end, the Commission is planning a frame work agreement with the Internet group Google and other companies. However, the proposal is still the subject of ongoing discussions.

With the release of ChatGPT, OpenAI has sparked the current hype about generative AI. It simulates human interaction and can create texts based on a few keywords. According to experts, this also increases the risk of disinformation campaigns.

Sam Altman, the CEO of OpenAI, the company behind ChatGPT, expressed his belief that the future of artificial intelligence involves finding new methods to create AI models beyond simply training them on existing knowledge.

Altman likened the growth of artificial intelligence to the dawn of agriculture or the development of machines in the industrial era. He emphasized that people will utilize these tools to innovate and shape the future we collectively inhabit.

However, individuals in various industries, particularly those in the arts and entertainment fields, do not share Altman’s optimism regarding the increasing sophistication of AI tools. There are concerns about the use of copyrighted material to train AI models and the proliferation of AI-generated disinformation such as deepfakes.

Altman acknowledged that it was “inevitable” for AI technology to be capable of more nefarious uses and expressed concerns about potential misuse of AI, including deepfakes, especially in the context of global elections.

OpenAI was also under scrutiny for its voice assistant Sky, which some online users noted sounded similar to the voice of Scarlett Johansson. OpenAI clarified that Sky’s voice was not an imitation of Johansson’s and belonged to a different actor hired by the company.

During a panel discussion, Altman and Airbnb co-founder and CEO Brian Chesky, who have been friends for over a decade, highlighted their strong relationship, which was instrumental in Altman’s reinstatement at OpenAI after he was fired.

OpenAI, a prominent AI startup, played a pivotal role in the development of generative AI technologies, including the launch of ChatGPT in 2022, which led to the proliferation of various AI tools such as hyperrealistic videos, humanlike music composition, and conversational chat agents.

Despite concerns about the potential implications of advancements in AI, Altman emphasized that even with the development of artificial general intelligence, these technologies would remain tools and not autonomous beings. He views the development of AI as a gradual evolution rather than a race and acknowledges the responsibility to get it right.

Sam Altman’s role in OpenAI’s Safety and Security Committee has raised concerns about its independence. As a result, Altman will no longer be part of the organization’s Safety and Security Committee, which aims to provide independent oversight of the AI models developed and deployed by the Microsoft-backed startup.

The committee was established in May 2024 to provide safety recommendations for the AI models developed and deployed by the startup backed by Microsoft. Concerns were raised about Altman leading the oversight body, suggesting that members might not be able to impartially assess the safety and security of its AI models.

With the CEO no longer in charge, the committee now includes two OpenAI board members – former NSA chief Paul Nakasone and Quora co-founder Adam D’Angelo – as well as Nicole Seligman, the former executive vice president at Sony, and Zico Kolter, director of the machine learning department at Carnegie Mellon University’s school of computer science.

According to OpenAI’s blog post published on Monday, September 16, “The Safety and Security Committee will receive briefings from company leadership on safety evaluations for major model releases, and will, along with the full board, oversee model launches, including having the authority to postpone a release until safety concerns are addressed.”

Upon the release of its new reasoning-based AI model o1, OpenAI stated that the safety committee had “examined the safety and security criteria used to assess OpenAI o1’s suitability for launch as well as the results of safety evaluations of OpenAI o1.”

The committee also completed its 90-day review of OpenAI’s processes and safeguards and provided the following recommendations to the AI firm:
- Establish independent governance for safety & security
- Enhance security measures
- Maintain transparency about OpenAI’s work
- Collaborate with external organizations
- Unify OpenAI’s safety frameworks for model development and monitoring
Before establishing the safety committee, both current and former employees of OpenAI had expressed concerns that the company was growing too rapidly to operate safely. Jan Leike, a former executive who left OpenAI along with chief scientist Ilya Sutskever, had posted on X that “OpenAI’s safety culture and processes have taken a backseat to shiny products.”

On May 16, 2023, Sam Altman, OpenAI’s charismatic, softly spoken, eternally optimistic billionaire CEO, and I appeared before the US Senate judiciary subcommittee meeting on AI oversight in Washington DC. At the time, AI was at the peak of its popularity, and Altman, then 38, was at the forefront of it all.

Hailing from St Louis, Missouri, Altman was the Stanford dropout who had risen to become the president of the highly successful Y Combinator startup incubator before the age of 30. A few months prior to the hearing, his company’s product ChatGPT had gained widespread attention.

Throughout the summer of 2023, Altman was treated like a celebrity, touring the world, meeting with prime ministers and presidents. US Senator Kyrsten Sinema praised him, saying, “I’ve never met anyone as smart as Sam… He’s an introvert and shy and humble… But… very good at forming relationships with people on the Hill and… can help folks in government understand AI.”

Flattering profiles at the time depicted the youthful Altman as genuine, talented, wealthy, and solely interested in advancing humanity. His frequent assertions that AI could revolutionize the global economy had world leaders eagerly anticipating it.

Gradually, I realized that I, the Senate, and ultimately the American people, had likely been deceived.

Senator Richard Blumenthal had summoned the two of us (and IBM’s Christina Montgomery) to Washington to discuss what should be done about AI, a “dual-use” technology with great promise but also the potential to cause great harm – from floods of misinformation to enabling the spread of new bioweapons. The focus was on AI policy and regulation. We pledged to tell the whole truth and nothing but the truth.

Altman represented one of the leading AI companies, while I was present as a scientist and author known for my skepticism about many things related to AI. I found Altman surprisingly engaging.

There were instances when he evaded questions (most notably Blumenthal’s “What are you most worried about?”, which I pressed Altman to answer more honestly), but overall, he seemed authentic, and I recall conveying this to the senators at the time. We both strongly advocated for AI regulation. However, little by little, I came to realize that I, the Senate, and ultimately the American people, had probably been manipulated.

In reality, I had always harbored some reservations about OpenAI. For example, the company’s publicity campaigns were often exaggerated and even deceptive, such as their elaborate demonstration of a robot “solving” a Rubik’s Cube that was later revealed to have special sensors inside. It received significant media attention, but ultimately led nowhere.

For years, the name OpenAI – which implied a commitment to openness about the science behind the company’s activities – had felt disingenuous, as it had become progressively less transparent over time.

The constant suggestion from the company that AGI (artificial general intelligence, AI that can at least match the cognitive abilities of any human) was just around the corner always seemed like unwarranted hype to me. However, in person, Altman was very impressive; I started to question whether I had been too critical of him before. Looking back, I realized that I had been too lenient.

I began to reconsider my opinion after receiving a tip about a small but revealing incident. During a Senate hearing, Altman portrayed himself as much more altruistic than he actually was. When Senator John Kennedy asked him, “OK. You make a lot of money. Do you?” Altman replied, “I make no… I get paid enough for health insurance. I have no equity in OpenAI,” and continued to elaborate, stating, “I’m doing this because I love it.” The senators were impressed by his response.

However, Altman wasn’t completely truthful. While he didn’t own any stock in OpenAI, he did own stock in Y Combinator, which in turn owned stock in OpenAI. This meant that Sam had an indirect stake in OpenAI, a fact acknowledged on OpenAI’s website. If that indirect stake were worth just 0.1% of the company’s value, which seems plausible, it would be worth nearly $100m.

This omission served as a warning sign. When the topic resurfaced, he could have rectified it, but he chose not to. People were drawn to his selfless image. (He even reinforced this image in an article with Fortune, claiming that he didn’t need equity with OpenAI because he had “enough money”.) Not long after that, I discovered that OpenAI had made a deal with a chip company in which Altman owned a stake. The selfless persona he projected began to seem insincere.

In hindsight, the discussion about money wasn’t the only thing from our time in the Senate that felt less than candid. The more significant issue was OpenAI’s stance on AI regulation. Publicly, Altman expressed support for it, but the reality was far more complex.

On one hand, perhaps a small part of Altman genuinely desired AI regulation. He often quoted Oppenheimer and acknowledged the serious risks that AI poses to humanity, likening it to nuclear weaponry. In his own words at the Senate (albeit after some prompting from me), he said, “Look, we have tried to be very clear about the magnitude of the risks here… My worst fears are that we cause significant – we, the field, the technology, the industry – cause significant harm to the world.”

However, behind closed doors, Altman’s lobbyists continued to push for weaker regulation of AI, or no regulation at all.

Presumably, Altman wouldn’t want to be remembered poorly. Yet behind closed doors, his lobbyists persistently lobbied for weaker regulation or none at all.

A month after the Senate hearing, it was revealed that OpenAI was working to soften the EU’s AI act. When Altman was dismissed by OpenAI in November 2023 for being “not consistently candid” with its board, I wasn’t entirely surprised.

At the time, few people supported the board’s decision to dismiss Altman. A large number of supporters rallied behind him, treating him like a saint. The well-known journalist Kara Swisher (known to be quite friendly with Altman) blocked me on Twitter simply for suggesting that the board might have been justified.

Altman handled the media adeptly. Five days later, with the support of OpenAI’s major investor, Microsoft, and a petition from employees backing him, he was reinstated.

However, much has changed since then. In recent months, concerns about Altman’s honesty have gone from being considered rebellious to being fashionable. Journalist Edward Zitron wrote that Altman was “a false prophet – a seedy grifter that uses his remarkable ability to impress and manipulate Silicon Valley’s elite.”

Ellen Huet of Bloomberg News, on the podcast Foundering, reached the conclusion that “when [Altman] says something, you cannot be sure that he actually means it.”

Paris Marx has cautioned against “Sam Altman’s self-serving vision.” AI pioneer Geoffrey Hinton recently questioned Altman’s motives. I myself wrote an essay called the Sam Altman Playbook, analyzing how he had managed to deceive so many people for so long, using a combination of hype and apparent humility.

Many factors have contributed to this loss of faith. For some, the tipping point was Altman’s interactions earlier this year with Scarlett Johansson, who explicitly asked him not to create a chatbot with her voice.

Altman proceeded to use a different voice actor, but one who was obviously similar to her in voice, and tweeted “Her” (a reference to a movie in which Johansson provided the voice for an AI). Johansson was furious.

The ScarJo incident highlighted a larger problem: major corporations like OpenAI claim that their models cannot function without being trained on all of the world’s intellectual property, but they have not fairly compensated many of the creators, such as artists and writers. Justine Bateman described this as “the largest theft in the history of the United States.”

Although OpenAI has repeatedly emphasized the importance of developing safety measures for AI, several key staff members focused on safety have recently left, stating that the company did not fulfill its promises. Jan Leike, a former OpenAI safety researcher, criticized the company for prioritizing flashy advancements over safety, a sentiment echoed by another former employee, William Saunders.

Co-founder Ilya Sutskever departed and launched a new venture called Safe Superintelligence, while former OpenAI employee Daniel Kokotajlo also expressed concerns that safety commitments were being disregarded. While social media has had negative impacts on society, the inadvertent development of problematic AI by OpenAI could be even more detrimental, as noted by Altman himself.

The disregard for safety exhibited by OpenAI is compounded by the company’s apparent efforts to silence its employees. In May, journalist Kelsey Piper uncovered documents revealing that the company could reclaim vested stock from former employees who did not agree to refrain from speaking negatively about the company, a practice that many industry insiders found alarming.

Subsequently, numerous former OpenAI employees signed a letter at righttowarn.ai requesting whistleblower protections, prompting the company to retract its decision to enforce these contracts.

Even the company’s board members felt deceived. In May, former OpenAI board member Helen Toner stated on the Ted AI Show podcast, “For years, Sam made it really difficult for the board… by withholding information, misrepresenting company events, and in some cases, outright lying to the board.”

By late May, negative publicity about OpenAI and its CEO had accumulated to the point where venture capitalist Matt Turck posted a cartoon on X: “days since the last easily avoidable OpenAI controversy: 0.”

There is a lot at stake. The way that AI is currently developing will have long-term implications. Altman’s decisions could significantly impact all of humanity, not just individual users, in enduring ways. OpenAI has acknowledged that its tools have been utilized by Russia and China to create disinformation, presumably to influence elections.

More advanced forms of AI, if developed, could pose even more serious risks. Despite the impact of social media on polarizing society and subtly influencing people’s beliefs, major AI companies could exacerbate these issues.

Moreover, generative AI, popularized by OpenAI, is having a substantial environmental impact in terms of electricity usage, emissions, and water consumption. As Bloomberg recently stated, “AI is already wreaking havoc on global power systems.” This impact could grow significantly as models continue to expand in size, which is the objective of all major players.

To a large extent, governments are relying on Altman’s assurances that AI will ultimately be beneficial, despite the lack of evidence so far, to justify the environmental costs.

I genuinely believe that if we continue on the current path, we will not achieve an AI that we can trust.

Meanwhile, OpenAI has taken a leading role, and Altman sits on the homeland security safety board. His counsel should be viewed with skepticism.

Altman may have briefly attempted to attract investors for a $7 trillion investment in infrastructure related to generative AI, which might end up being a significant waste of resources that could be better utilized elsewhere if, as many suspect, generative AI is not the right path to AGI [artificial general intelligence].

Overestimating current AI could potentially lead to conflicts. For example, the US-China “chip war” concerning export controls, where the US is restricting the export of crucial GPU chips designed by Nvidia and manufactured in Taiwan, is affecting China’s AI progress and escalating tensions between the two nations.

The chip battle is largely based on the belief that AI will continue to advance exponentially, despite data indicating that current approaches may have reached a point of diminishing returns.

Altman may have initially had good intentions. Perhaps he genuinely aimed to protect the world from AI threats and guide AI for positive purposes. However, greed might have taken over, as is often the case.

Unfortunately, many other AI companies appear to be following the same path of hype and cutting corners as Altman. Anthropic, formed by a group of OpenAI ex-employees concerned about the lack of focus on AI safety, seems to be increasingly competing directly with its parent company.

The billion-dollar startup Perplexity also appears to be a lesson in greed, using data it should not be using. Meanwhile, Microsoft shifted from advocating “responsible AI” to rapidly releasing products with significant issues, pressuring Google to do the same. Money and power are corrupting AI, much like they corrupted social media.

We cannot rely on large privately held AI startups to self-govern in ethical and transparent ways. If we cannot trust them to govern themselves, we certainly should not allow them to govern the world.

I sincerely believe that we will not achieve trustworthy AI if we continue on the current path. Apart from the corrupting influence of power and money, there is also a significant technical issue: large language models, the fundamental technique of generative AI, are unlikely to be safe. They are inherently stubborn and opaque – essentially “black boxes” that we can never fully control.

The statistical techniques behind these models can achieve remarkable feats, such as accelerating computer programming and creating believable interactive characters resembling deceased loved ones or historical figures. However, such black boxes have never been reliable and are therefore an unsuitable basis for AI that we can entrust with our lives and infrastructure.

Nonetheless, I do not advocate for abandoning AI. Developing better AI for fields like medicine, material science, and climate science could truly revolutionize the world. Generative AI may not be the solution, but a future form of AI yet to be developed might be.

Ironically, the biggest threat to AI today could be the AI companies themselves; their unethical behavior and exaggerated promises are turning many people away. Many are ready for the government to take a more active role. According to a June survey by the Artificial Intelligence Policy Institute, 80% of American voters prefer “regulation of AI that mandates safety measures and government oversight of AI labs instead of allowing AI companies to self-regulate.”

To achieve trustworthy AI, I have long advocated for an international effort similar to Cern’s high-energy physics consortium. The time for that is now. Such an initiative, focused on AI safety and reliability rather than profit, and on developing a new set of AI techniques that belong to humanity rather than just a few greedy companies, could be transformative.

Furthermore, citizens need to voice their opinions and demand AI that benefits the majority, not just a select few. One thing I can guarantee is that we will not achieve the promised potential of AI if we leave everything in the hands of Silicon Valley. Tech leaders have been misleading for decades. Why should we expect Sam Altman, last seen driving a $4 million Koenigsegg supercar around Napa Valley, to be any different?

When did OpenAI start?

OpenAI embarked on its groundbreaking journey on December 11, 2015, as a response to the potential dominance of AI by large tech companies.

Who are the current owners of OpenAI?

During its early stages, OpenAI received substantial support from influential figures in the industry, including contributions from Elon Musk and Peter Thiel.

As the company evolved, Elon Musk decided to step down from the board in 2018 to avoid potential conflicts with his other ventures like Tesla and SpaceX.

Due to its ambitious goals and financial requirements, OpenAI transitioned from a nonprofit to a “capped-profit” for-profit entity in 2019, with a significant $1 billion investment from Microsoft.

Ownership of OpenAI is divided among Microsoft (49%), other stakeholders (49%), and the original OpenAI non-profit foundation, which maintains its autonomy.

Other stakeholders in OpenAI include a16z, Sequoia, Tigers Global, and Founders Fund.

OpenAI Inc. functions as the overarching non-profit umbrella, while its for-profit activities are managed by OpenAI LP.

Is OpenAI a publicly traded company?

Despite its significant presence in the AI field, OpenAI is a private company and is not subject to the strict regulations and quarterly pressures faced by public companies.

However, there is considerable demand for OpenAI stock, so a public offering cannot be ruled out in the future.

Conflicts within the OpenAI Board

Elon Musk Sues OpenAI for ‘Placing Profit Above Humanity’

In late February 2024, Elon Musk, who co-founded OpenAI in 2015, filed a lawsuit against OpenAI, alleging that the company had shifted its focus from creating artificial intelligence for the benefit of humanity to pursuing profit.

Musk claims that OpenAI, which was established as a not-for-profit organization with the goal of developing artificial general intelligence, has become a closed-source subsidiary of Microsoft, focusing on maximizing profits for the company.

Musk’s lawsuit seeks to compel OpenAI to adhere to its founding agreement and return to its mission of developing AGI for the benefit of humanity.

In response to Musk’s claims, OpenAI released an open letter stating that Musk had been involved in discussions about creating a for-profit entity in 2017 and had sought majority equity and control over the board and CEO position.

Elon Musk decided to leave OpenAI and started his own AGI competitor named xAI within Tesla.

Sam Altman’s Unexpected Departure from OpenAI

On November 17, 2023, Sam Altman was unexpectedly removed from his position as CEO of OpenAI.

Mira Murati, the company’s chief technology officer, assumed the role of interim CEO, and Emmett Shear, the former CEO of Twitch, was appointed as the new CEO.

Microsoft CEO Satya Nadella offered Altman a position to lead an internal AI division at Microsoft, which Altman accepted, and OpenAI’s president Greg Brockman also transitioned to a role at Microsoft.

However, just four days later, Sam Altman resumed his position as CEO of OpenAI, despite having accepted a role at Microsoft.

OpenAI’s founder and CEO, Sam Altman, recently saw his net worth reach $2 billion according to the Bloomberg Billionaire Index. However, this figure does not reflect any financial gains from the AI company he leads.

This marks the first time the index has assessed the wealth of the 38-year-old, who has become synonymous with artificial intelligence as the CEO of OpenAI, which was recently valued at $86 billion.

According to a report by Bloomberg, Altman has consistently stated that he does not hold any equity in the organization. The report also indicated that a significant portion of his observable wealth comes from a network of venture capital funds and investments in startups.

Moreover, his wealth is expected to increase with the upcoming initial public offering of Reddit, where he stands as one of the largest shareholders.

In related news, Tesla’s CEO, Elon Musk, has filed a lawsuit against OpenAI and Sam Altman, accusing them of violating contractual agreements made when Musk helped establish the ChatGPT developer in 2015.

A lawsuit submitted on Thursday in San Francisco claims that Altman, along with OpenAI’s co-founder Greg Brockman, originally approached Musk to develop an open-source model.

The lawsuit further stated that the open-source initiative promised to advance artificial intelligence technology for the “benefit of humanity.”

In the legal filing, Musk alleged that the focus on profit by the Microsoft-backed company violates that agreement.

It is important to mention that Musk co-founded OpenAI in 2015 but resigned from its board in 2018. Subsequently, in October 2022, Musk acquired Twitter for $44 billion.

OpenAI’s ChatGPT became the fastest-growing software application globally within six months of its launch in November 2022.

Moreover, ChatGPT triggered the development of competing chatbots from companies such as Microsoft, Alphabet, and various startups that capitalized on the excitement to secure billions in funding.

Since ChatGPT’s introduction, many companies have started utilizing its capabilities for diverse tasks. This includes document summarization, coding, and igniting a competitive race among major tech firms to release their own generative AI products.

Although OpenAI is currently valued at $157 billion, it still faces challenges ahead

Recently, OpenAI completed the most lucrative funding round in Silicon Valley’s history. The next step is to successfully navigate a highly competitive AI landscape.

Even though Sam Altman’s company has solidified its leading position in the generative AI boom by achieving a new $157 billion valuation after securing $6.6 billion in fresh capital from prominent investors, its top position is not assured.

Since the launch of ChatGPT in late 2022, it has become evident that their mission to create large language models that can rival human intelligence will involve substantial costs necessitating extensive resources.

Though Altman’s company now casts a significant influence over the industry with its new valuation of $157 billion, numerous competitors are vying for capital and resources, making the startup’s path to profitability more complex.

Thus, while OpenAI has a moment to commend, the situation will soon reveal how strong its competitive advantage is and whether a severe wave of consolidation is imminent in Silicon Valley’s booming industry.

While OpenAI’s recent valuation and capital influx represent enormous amounts that any founder in Silicon Valley would envy, indications suggest that Altman remains somewhat apprehensive.

As per a Financial Times report about the fundraising, Altman’s nearly nine-year-old venture urged its new investors—a group led by Thrive Capital, which includes Nvidia, SoftBank, and Microsoft—to refrain from funding rival companies, of which there are many.

Anthropic and Mistral, both valued in the billions, are aiming to challenge OpenAI. Additionally, Musk’s xAI and Safe Superintelligence (SSI), a startup founded in June by Ilya Sutskever, a former chief scientist at OpenAI who previously attempted a coup against his ex-boss, are also in the mix.

“For the main model developers, these mega-funding rounds are becoming standard as the expenses for training the largest models are soaring into the hundreds of millions of dollars,” remarked Nathan Benaich, founder and partner at Air Street Capital, an investment firm.

Several significant factors indicate that OpenAI cannot afford to be complacent.

For starters, the expenses associated with delivering groundbreaking advancements in generative AI are projected to escalate. Dario Amodei, CEO of Anthropic, noted earlier this year that he anticipates training expenses for AI models could exceed $10 billion by 2026 and potentially reach $100 billion afterward.

OpenAI itself might face training costs surpassing $3 billion annually, as previously estimated by The Information. Training GPT-4o, for instance, costs around $100 million, but this figure is expected to increase based on the complexity of future AI models.

A portion of the expenses is fueled by the acquisition of powerful chips, referred to as GPUs, primarily sourced from Jensen Huang’s Nvidia, to establish clusters in data centers. These chips are crucial for supplying the computational strength necessary to operate large language models (LLMs).

The competition for talent has been intense in this current wave, as AI laboratories strive to gain an advantage over their rivals, prompting them to present ever more extravagant compensation packages.

“Benaich remarked to BI, “These expenses are set to escalate as firms continue to invest heavily to compete for often slight performance improvements over their rivals. This competition lacks clear historical comparisons, largely due to the staggering capital expenditure requirements and the absence of a straightforward path to profitability.”

Although OpenAI’s newfound capital will assist in financing some of the more costly aspects of its operations, it isn’t exactly in a strong position at this moment. A report from The New York Times last week indicated that the leading AI laboratory globally is poised to finish the year with a $5 billion deficit.

Additionally, OpenAI’s rumored push for exclusivity among its investors may have potential downsides. Benaich characterized this approach as “uncommon” but also as a representation of how OpenAI views its own clout in the market.

“This is also a daring strategy that may attract unwanted scrutiny from regulatory bodies,” he added.

For experts in the industry, this situation poses questions about the long-term sustainability of such practices.

Investors foresee some degree of consolidation approaching.

As OpenAI solidifies its role as the leading player in the industry, investors expect some consolidation among startups focusing on foundational models in the upcoming year.

LLM startups require continuous access to substantial capital, but not everyone can secure the same inflow of funds as OpenAI. With Microsoft acquiring Inflection.ai and Google similarly attracting the founding team of Character.ai, investors anticipate more acquisitions of this nature in the near future.

“This is a competition for capital as well, and ultimately only financial backers like sovereign wealth funds will be capable of providing the necessary capital for these LLM startups,” a European growth-stage venture capitalist mentioned to BI.

When funding becomes scarce, established giants, including major tech companies, might acquire smaller focused companies. These smaller firms have access to a vast array of proprietary data for training their models.

Venture capitalists also predict a more grounded approach to investing in LLM leaders at inflated valuations. “Many other firms are raising funds based on aspiration rather than substance, and I believe we will begin to witness a certain rationalization in that area,” another growth-stage VC informed BI, noting that “the overheated excitement surrounding AI is likely to temper next year.”

“You don’t require 50 foundational model enterprises — it’s more likely that you’ll end up with two or four,” he stated.

He added that those companies which endure will be the ones that effectively cater to consumer needs. “You might see Amazon, Anthropic, OpenAI, Meta, and Google, but I struggle to envision many others existing.”

OpenAI has successfully secured $6.6 billion in a significant funding round that places a valuation of $157 billion on the startup, placing it in a small group of tech startups with extraordinarily high private valuations.

This deal, which roughly doubles OpenAI’s valuation from just this past February, highlights the intense expectations investors have for the generative AI surge that OpenAI catalyzed with the launch of ChatGPT in 2022.

“The new funds will enable us to strengthen our position in leading-edge AI research, enhance our computing power, and continue developing tools that assist people in tackling challenging problems,” OpenAI stated in its announcement regarding the deal on Wednesday.

The funding arrives as the nine-year-old AI startup, helmed by CEO Sam Altman, confronts rising competition from companies like Google, Meta, and other AI startups, and during a time when OpenAI is navigating its own transitions — most famously marked by a boardroom incident last year that saw Altman briefly ousted and then reinstated within five days.

Since that time, the firm has faced a series of significant leadership exits as it tries to shift from its origins as a nonprofit research entity to a producer of commercial products that can take advantage of the booming AI sector. Recently, OpenAI’s chief technology officer Mira Murati unexpectedly stepped down to “create the time and space for my own exploration.” Moreover, as recently reported by Fortune, some insiders have expressed concerns that the company’s focus on safety may have been compromised in the rush to release new products ahead of competitors.

Despite the internal upheaval, investors seemed eager to gain a stake in the startup.

OpenAI did not reveal the identities of its investors, but Thrive Capital confirmed via email to Fortune that they had invested and led this latest funding round. According to Bloomberg, which first shared news of the deal, Khosla Ventures, Altimeter Capital, Fidelity, SoftBank, and the Abu Dhabi-based MGX also joined in, along with AI chip manufacturer Nvidia and Microsoft, which had previously invested $13 billion.

OpenAI has reported that ChatGPT is used by over 250 million individuals weekly

With this funding, OpenAI solidifies its position as one of the most valuable startups globally, following TikTok parent company ByteDance, valued at $225 billion, and SpaceX, led by Elon Musk, with a valuation of $200 billion, according to CB Insights’ rankings of tech company valuations.

On Wednesday, OpenAI announced that more than 250 million people worldwide engage with ChatGPT weekly.

While the company does not share its financial outcomes, the New York Times has indicated that OpenAI’s monthly earnings reached $300 million in August and anticipates generating $11.6 billion in revenue in the coming year.

With a new valuation of $157 billion after funding, investors seem to be assessing the company at 13 times its expected revenue for next year.

In comparison, Google’s parent company, Alphabet, is traded on the stock market at 5.3 times its predicted revenue for next year, while Nvidia is evaluated at approximately 16 times projected revenue.

On Wednesday, OpenAI referenced its foundational principles, emphasizing that it is “making strides toward our goal of ensuring that artificial general intelligence serves the entire human race.”

Artificial general intelligence, or AGI, remains a theoretical concept of an AI system capable of performing tasks as well as or even better than humans.

The potential risks associated with AGI were part of the rationale behind OpenAI’s establishment in 2015, as Altman, Elon Musk, and the other co-founders aimed to create a counterbalance to Google’s DeepMind, which they were concerned would develop AGI driven solely by commercial motives.

Musk, who departed from OpenAI, has criticized the organization for straying from its original purpose, even as he has ventured into his own AI enterprise, xAI.

The valuation of OpenAI has nearly doubled since earlier this year when it arranged a tender offer allowing employees to sell a portion of their shares to private investors, valuing the company at about $80 billion.
November 8, 2024
The debate around artist compensation in AI art, and some possible solutions to the problem
Artificial intelligence uses computer programs to make large scale use of products of human creativity. Artists, graphic designers and authors ask themselves: Is that fair?

The new image and speech programs, especially ChatGPT, have quickly turned the world of so-called knowledge workers upside down. And that was exactly the intention of the company Open AI. ChatGPT is intended to”help” creative people to compose songs, write screenplays or imitate the styles of writers, explained Open AI boss Sam Altman. And it can make all of this work cheaper and thus replace it: “The cost of intelligence, of intelligent work, will tend towards zero. I hope that will happen, ” said Altman in a podcast.

Text, images or music – previously the work of human hands or minds – can now be produced automatically and in series by AI, soon for free. The triumph of artificial intelligence could make many jobs redundant. In addition, AI image generators currently use material that they store in their databases from all corners of the Internet. They do not take into account images that are protected by copyright.

“Horse-drawn carriage drivers also thought cars were bad”

“There are enough artists who have been told: Yes, thank you for the offer, we’ve run your daily rate through the system. We’ve found that we can generate everything more cheaply with Midjourney,” says graphic designer and publisher Spiridon Giannakis. He calls for strict regulation and for AI companies to have to compensate artists.

Richard Socher is considered the most influential German in the artificial intelligence industry. In Silicon Valley, he founded the AIsearch engine You.com – a competitor to ChatGPT and Google. Graphic designers have to accept that the world is changing, he says in an interview with the ARD magazine Panorama : “Horse-drawn carriage drivers also thought it was bad that cars could drive automatically and that you no longer needed a carriage driver. The same applies if you are now an illustrator.”

His company offers AI-generated images – but he doesn’t want to compensate artists for them. “Dali painted the clock in a slightly outdated way. And if anyone ever says: Oh, I want to have an outdated object in my picture, then Dali comes along and says that it was influenced by me and now you have to pay me maybe five euros per pixel. That doesn’t make sense.”He can understand the creatives. “If an artist is currently making money from it, of course he doesn’t want automation,” says Socher. Everyone just wants to make as much money as possible.

Billion-dollar corporations benefit

The reason why AI produces surprisingly good results isbecause the language programs have been fed billions of parameters, especially the content of those that could then be replaced by the AI. Companies are thus absorbing the world’s knowledge and skills and copying styles without paying or acknowledging the creatives. Everything AI does is fed by the works of countless people made available on the Internet.

Creatives complain that this is cynical and threatens their existence, because the “art” generators are trained with their images. “Who is currently profiting from artificial intelligence? Is it us or those who have founded billion-dollar companies on the backs of the people whose data was fed into it? That’s not fair,” says graphic designer Giannakis. In every conversation he has with artists, there is great concern.

You.com founder Socher has been working in Silicon Valley for ten years. He is surprised that Europeans are so skeptical about the new technology. Things are completely different in California: “When a new technology comes along there, I see hundreds of my friends, especially in Silicon Valley, saying: Wow, how can I use this now? And maybe I can open a start-up there that uses this new technology to make something even more productive, even more efficient. In Germany, the attitude is initially: Whatcould go wrong with this? Job loss? How do we have to regulate this before it even works properly?”

Texts as raw material

Former journalist Michael Keusgen founded the company Ella.The Cologne-based start-up fed its language models with massive amounts of text data: with essays, specialist books, but also with fiction – texts as raw material. However, Keusgen bought the rights for this. In this way, he wants to revolutionize the media industry, especially in print and online editorial departments.

“We are currently producing paraphrased texts and will be writing more and more texts. But when it comes to facts, the human component is essential,” explains Keusgen. There has to be an editor who does the proof reading at the end to check it.

Its language models work like all major AI programs: they calculate, based on statistical probability, which word or sentence might come next – and the results don’t always make sense. So you can’t expect the AI to always tell the truth, because it can’t distinguish fiction from reality. The answers can seem convincing, even if they aren’t based on facts.

Unsuitable for facts

Computer scientist Katharina Zweig therefore advises against using AI in journalism: “I believe that if you use AI systems to write texts whose factual content you cannot verify yourself, then you are using these machines completely wrongly. They have not been trained for this.”

That’s what went wrong with Open AI. It’s a dangerous misunderstanding that ChatGPT can be used to explain quantum computing to six year olds, for example. That’s why she recommends: “Don’t use it for texts whose factual content you can’t check yourself.”

Cost of Developing AI Software in 2024

In today’s world, artificial intelligence (AI) stands as one of the most successful innovations. The concept of creating AI software is at the forefront of every business owner’s mind, and numerous online businesses are already integrating it. This represents a significant opportunity to enhance business operations and increase revenue and customer base.

AI software is widely embraced by customers and technology enthusiasts worldwide, regardless of the target audience.

We are currently in a rapidly evolving tech landscape where AI is poised to continue its dominance in 2024, revolutionizing business processes and reducing time spent on repetitive tasks.

As companies strive to fully leverage the power of AI, a crucial question arises: “What is the Cost of Developing AI Software in 2024?”

This article aims to explore the total cost of developing AI software in 2024.

Estimated Cost of Developing AI Software in 2024

The cost of developing AI software can vary depending on the specific requirements. As a rough estimate, the cost of AI software development can reach up to $400,000. It’s important to note that this is just an estimate.

To gain a better understanding of the cost, it’s essential to carefully assess the project requirements and consider various factors such as project type and development, as these can significantly impact the cost of AI software development.

The following provides a rough estimate for different types of AI projects:
- Small-scale AI project: Estimated cost ranges from $10,000 to $100,000.
- Medium-scale AI project: Estimated cost ranges from $100,000 to $500,000.
- Large-scale AI project: Complex applications like healthcare diagnostics, autonomous vehicles, and advanced natural language processing systems can cost anywhere from $500,000 to $900,000.
- Enterprise-level AI project: Organizations with extensive AI initiatives may invest over $900,000.
For an accurate software development cost estimation, it’s recommended to consult with an AI development company.

When consulting with professionals, it’s crucial to thoroughly outline all project details to avoid any unexpected additional costs from the development team.

Key Factors Influencing the Cost of AI Software Development

Project Type

The first step is determining whether a custom or off-the-shelf AI solution is needed. Custom solutions involve building and training AI from scratch to meet specific objectives, while off-the-shelf AI consists of pre-structured algorithms tailored for specific purposes.

Successful AI solutions must meet business expectations and requirements, requiring time and effort from ideation to deployment. Custom AI development costs can range from $5,000 to $150,000.

Data Requirements

AI heavily relies on data, and the amount, quality, and availability of data for training and refining AI models directly impacts costs. Collecting, refining, and organizing data requires time and resources, increasing overall project costs. Projects requiring a large amount of high-quality data can also affect infrastructure costs.

Development of Advanced AI Technologies

AI development depends on high-speed hardware, specialized software, and computing resources. Considering the cost impact of cloud-based solutions versus on-premises hardware is crucial. Infrastructure costs may increase for advanced AI projects due to the demand for computing power.

Integration of AI Software Features

AI solutions are distinguished by their features, some of which may be necessary while others may not be. For instance, natural language processing is essential for generating text or answering questions, and deep learning is part of machine learning. Speech and image recognition may also be integrated. The implementation of these features significantly impacts the development cost of AI, and industry-trusted features add to the overall cost.

Hardware Costs

If you develop AI software internally or hire a third party to do it, you will incur hardware expenses. When you hire a company to create AI software, the cost typically encompasses more than just software development. They are focused solely on software development. However, the AI algorithms require computing power to process and analyze data.

To support this process, a powerful and specialized infrastructure is needed to handle large computations. Consequently, you will need to allocate funds for hardware and AI software development.

Development team

The team involved in development is another important factor that impacts development costs. Select a team that provides AI & ML Services. Small businesses might spend upwards of $320,000 annually on their AI development team.

AI development teams have several essential roles to fulfill. Typically, team members include data scientists, machine learning engineers, artificial intelligence developers, and software developers. The cost of each member depends on their skills and experience. Additionally, the number of team members assigned to your project also affects the cost.

Maintenance and management

The management of AI software can be handled internally or outsourced. While outsourced teams may be more expensive, they eliminate in-house costs such as employee salaries.

Building an AI is one thing, but maintaining it is another. While it may be possible to train the algorithm to process data and perform computations, the team will be responsible for maintaining the AI and ensuring it meets business requirements. This ensures that its performance and efficiency are optimized.

Duration of the project

Finally, the cost of AI development is influenced by the duration of the project. All the factors mentioned above will impact the duration. An AI developed as a basic version will be less expensive and require less time than one developed as an MVP.

Whether in-house or outsourced, a provider of ML services that works for longer durations will need to dedicate more time and effort, resulting in a higher cost.

Conclusion

Developing Artificial Intelligence Software is a significant investment for transforming and automating business operations. The cost of building the software in 2024 can vary based on factors such as project type, development team, and more.

It is highly recommended to engage a professional AI development service provider to deliver a top-class AI solution that aligns with your business needs.

How much does AI cost?

The ITRex team estimates that you would spend a minimum of $50,000 on an MVP version of an AI solution, with the cost of artificial intelligence increasing in line with its complexity and supported use cases.

It is important to note that the above price applies only to the artificial intelligence component of your system; the efforts required to create custom web and mobile applications supporting its logic will be billed separately.

However, this does not prevent your company from implementing AI on a smaller scale and budget.

There are numerous ways to implement AI in business, from acquiring off-the-shelf call center chatbots to building a custom self-service BI solution that sources data from various enterprise systems. Therefore, the costs of artificial intelligence will vary depending on the approach and type of solution chosen.

For the purposes of this article, we will focus on customized and fully custom AI solutions. As an AI consulting company, ITRex will help you determine the factors that influence their development, enhancement, and maintenance costs.

Furthermore, our AI developers will provide rough estimates for several artificial intelligence projects from our portfolio, as well as advice for approaching your AI pilot and maximizing ROI.

Let’s get started!

What are the top 5 factors behind AI cost?

The type of software you intend to build. Artificial intelligence is a broad term that encompasses any device or application that makes decisions based on the information it processes, thus emulating human intelligence.

Voice assistants that understand natural language queries, security cameras that identify individuals in live video footage, and expert systems that detect cancerous tumors in CT scans all fall under the umbrella of artificial intelligence. However, their complexity, performance requirements, and consequently, costs, vary greatly.

The level of intelligence you aim to achieve. When discussing AI, people often envision robots from Boston Dynamics and holographic avatars from Blade Runner 2049.

In reality, most business AI solutions can be classified as narrow artificial intelligence, meaning they are programmed to perform specific tasks, such as recognizing text in PDF files and converting them into editable documents.

To be truly intelligent, AI algorithms should be able to uncover patterns in data with minimal human intervention, assess the probability or improbability of an event, justify their assumptions, continually process new data, and learn from it.

The quantity and quality of data you will input into your system is crucial. The effectiveness of artificial intelligence is directly linked to the data it has been trained on, and the more data algorithms process, the better they become.

The existence of pre-trained AI development tools, such as large language models (LLMs), makes the training process much easier. Some off-the-shelf solutions, like ChatGPT or DALL·E 3, can even be used without further customization.

However, the most optimal results are achieved by fine-tuning algorithms with unique data specific to your company. This data can be organized, stored in relational database management systems (RDBMs), or unstructured, like emails, images, and videos, which are typically bulk-uploaded to data lakes.

Regarding the cost of AI, working with structured data is more cost-effective, especially when dealing with a large quantity of information to enhance algorithm accuracy. With unstructured data, additional efforts are required to organize and label it, and software engineers need to establish a complete infrastructure to ensure continuous data flow within the system components. In some cases, such as training AI-powered medical imaging solutions, obtaining data can be challenging due to privacy or security concerns.

To overcome this obstacle, AI engineers may expand the size of a limited dataset, reuse existing classification algorithms, or create synthetic data for model training using generative AI solutions. These operations are likely to increase the cost of developing an AI program.

The level of accuracy you aim to achieve with your algorithm is crucial. The accuracy of your AI solution and its predictions is directly dependent on the type of application and the requirements you set for it. For example, a customer support chatbot is expected to handle up to 60% of routine user queries; for complex issues, human specialists are available.

Conversely, a pilotless delivery drone transporting blood and human organs must maneuver around objects with precise accuracy, relying on advanced computer vision algorithms. Higher accuracy and reliability of AI predictions directly impact the project’s longevity and increase the cost of AI development.

It’s worth noting that AI algorithms will continue to learn from new data as they work alongside human specialists, which may entail additional training and maintenance expenses.

The complexity of the AI solution you’re developing is also a key factor. Artificial intelligence is the core of a technology system that processes data for your business app and presents insights to users, including those without a technical background. When considering the cost of artificial intelligence, the cost of developing the actual software should be taken into account.

This includes a cloud-based back end, ETL/streaming tools, APIs for internal and external application integration, and some form of user interface, such as a cloud dashboard, mobile app, or voice assistant.

Simple AI, like the customer support chatbots mentioned earlier, may reside within a corporate messenger and does not require a complex infrastructure. On the other hand, AI-powered data ecosystems providing a comprehensive view of your company’s operations pose a different challenge.

Additional challenges in AI implementation arise when scaling your intelligent system from individual use cases to company-wide deployment. This is why only 53% of enterprise AI projects make it from prototypes to production.

Regarding failures, it should be noted that only a small fraction of AI projects (Gartner believes it’s 20%; VentureBeat is even less optimistic) actually deliver on their promise. Several factors contribute to such a high failure rate, including a lack of collaboration between data scientists and software engineers, limited or low-quality training data, and the absence of a company-wide data strategy.

Most failed AI projects are described as “moonshots”—overly ambitious endeavors led by idealistic data scientists and CIOs seeking to “completely change the way our company has been operating for decades.” Such projects may take a long time to complete, and it’s natural that, at some point, a company’s C-suite stops investing in a project without seeing real value.

How much does AI cost? The following examples from the ITRex portfolio may give you an idea:

Project 1: AI-powered telemedicine solution

A healthcare technology company approached ITRex to enhance a telehealth system, which is implemented in various hospitals across the USA, by adding video recording capabilities.

The latest version of the system would enable healthcare providers to utilize facial recognition and natural language processing technologies to analyze videos recorded during consultations, potentially enhancing doctor-patient interactions.

During the exploratory phase, we eliminated potential technological obstacles and chose the best tools for the project, primarily Python and the related frameworks and SDKs for speech recognition and analysis. The client opted for the speech-to-text functionality only for the initial version of the telemedicine system, with no user-facing components expected to be included.

The solution performs linguistic analysis of video recordings to identify potential changes in communication style that could provide insight into patients’ well-being and assist physicians in devising better treatment plans.

The estimated cost for a basic version of a video/speech analysis AI platform is $36,000 to $56,000.

Project 2: A smart recommendation engine

An entrepreneur wanted to incorporate AI capabilities into a B2C platform that connects users with local service providers. The client’s concept involved replacing complex search filters with advanced machine learning algorithms that would analyze input text and generate a list of service providers matching a user’s query.

We chose Amazon Personalize as the primary technology stack for the AI component of the project. In addition to offering personalized recommendations based on user queries, the recommendation engine comes with a fully managed cloud infrastructure for training, deploying, and hosting ML models. The backend of the system would be developed in Python, while user data would be securely stored in the cloud (Amazon S3).

The estimated cost for developing, testing, and deploying a similar artificial intelligence platform (MVP) ranges from $20,000 to $35,000.

Project 3: An AI-powered art generator

A well-known visual artist approached ITRex to develop a generative AI solution that would create new paintings based on his own works and the works of other inspiring artists. The client aimed to build a minimum viable product (MVP) version of the system over several weeks to showcase at an exhibition.

The ITRex team proposed creating a neural network based on Python frameworks (PyTorch, TensorFlow) to analyze abstract paintings, learn the artist’s distinctive style, generate similar images, and showcase them on the artist’s official website.

For the MVP version, we recommended using a 1000 x 1000 image resolution similar to Instagram and deploying the AI solution locally, with the option to migrate the system to the cloud in the future.

The estimated cost for building an MVP version of an artificial intelligence system like this could range from $19,000 to $34,000, depending on factors such as the type of training data and image resolution.

If your company is considering developing a generative AI solution, take a look at our guide on Gen AI costs. The article outlines various approaches to implementing generative AI, including using commercially available tools as is and retraining open-source models. Additionally, we suggest reading our blog post on machine learning implementation costs.

How to reduce AI costs — and start benefiting from artificial intelligence ASAP

According to a recent Forbes Technology Council article, the development and deployment of an AI solution will ultimately cost your company 15 times more than you anticipated if you do not have an efficiently built data ecosystem in place.

Higher AI development costs typically arise from significant infrastructure optimization, data integration, security, and artificial intelligence management and control efforts.

However, you can minimize these expenses by thoroughly planning your project and starting small while keeping the bigger picture in mind. You can also use pre-trained foundational AI models to expedite your project or experiment with artificial intelligence.

To help you develop an artificial intelligence system at a lower cost and begin reaping its benefits from the outset, the ITRex team has prepared a comprehensive AI development and implementation guide. The primary concept revolves around taking an agile approach, as it might be challenging to capture all the requirements for a custom AI solution or come up with a realistic artificial intelligence cost estimation at the beginning of your journey.

Another advantage of this approach is that it enables you to see a significant ROI early on, which can help secure buy-in from your company’s C-suite and secure further funding.

Collect feedback from stakeholders. Before starting to develop an AI system, it is suggested to consult with internal and external stakeholders to identify the key processes and decision flows that can be supplemented or automated with AI.

Identify the most important use cases. In this step, use a product prioritization framework (e.g., MoSCoW, RICE, or Kano) to choose business cases that will provide the most value during the interim period and serve as a basis for further AI implementations.

Choose the best technology stack. To build a vendor-agnostic solution and reduce overall AI development costs, use a mix of custom-made, open-source, and off-the-shelf components (for example, plug-and-play facial recognition engines, API-driven voice assistants, and cloud-based services supporting the creation and training of AI algorithms).

Pay special attention to UI/UX design: your future AI system should have a user-friendly interface that allows stakeholders to ask artificial intelligence questions, get instant insights, or automate tasks without seeking assistance from your IT department.

Prepare data for AI-driven analysis. To help algorithms understand your business data, it is crucial to gather information, assess its quantity and quality, and bring it into a unified format. There are several data collection, preparation, and normalization techniques that can be applied. More information can be found in our blog post on data preparation for machine learning.

Remember that identifying the right data and thoroughly preparing it for model training is crucial to reduce the cost of artificial intelligence while developing a system that produces consistent results.

Create a minimum viable product (MVP) of your AI system. Building an MVP supporting the essential use cases is one of AI development best practices. With an MVP, you can assess the feasibility of your concept, identify areas for algorithm improvement, and start scaling the system across different use cases and departments.

Do not confuse an MVP with an AI proof of concept (PoC); the latter validates your idea and is intended for internal use only. However, it’s often advisable to begin your AI journey with a proof of concept to test the feasibility of your idea and eliminate technology barriers early on.

Treat AI implementation as a continuous process. When you start using artificial intelligence, perfect results may not be immediate. As your AI system consumes new information under the supervision of human specialists, it will provide more accurate predictions and become more autonomous.

It is important to continue gathering feedback from your company’s stakeholders, making the necessary changes to the system, and repeating the steps described above when introducing new features and use cases. This will not only allow you to optimize the AI development cost but also help solve the artificial intelligence scalability problem.

Ultimately, how much does artificial intelligence cost?

Though estimating the cost of creating and implementing an artificial intelligence application without delving into your project’s details is difficult, you might spend around $50,000 on a very basic version of the custom system you’re looking to build. However, you can still initiate the process with a smaller budget, especially if you’re considering a PoC or using pre-trained ML models or plug-and-play services.

Is it worth it?

By 2030, artificial intelligence could contribute up to $15.7 trillion to the global economy, with increased productivity and automation driving the majority of this sum.

Currently, the AI revolution is still in its early stages. While some countries, industries, and companies might be better prepared for the disruption (meaning they have the necessary data and IT infrastructure in place to create and deploy custom AI solutions at scale), the competitive advantage is elusive since there is an opportunity for every business to transform the way they work and lead the AI race. And your company is no exception.

How Much Does it Cost to Build an AI System?

Building an AI system can be a transformative move for businesses. However, it involves various costs that can vary greatly depending on the type of business and the complexity of the AI system.

Based on my research and experience, I will outline the costs involved in building an AI system for different types of businesses: small businesses, medium-sized enterprises, and large corporations. I will also provide insights into the factors affecting these costs and some statistics to support the discussion.

AI Costing for Small Businesses

Small businesses often have limited budgets and resources. According to my research, the cost to build an AI system for small businesses can range from $10,000 to $50,000. Several factors influence this cost.

AI Solution Type: The cost is significantly influenced by the type of AI solution. For example, a basic chatbot or recommendation engine will be cheaper than a complex predictive analytics system.

Data Collection and Preparation: Small businesses may need to allocate funds for gathering and preparing data. This may involve expenses related to data cleaning, data labeling, and data storage.

Development and Deployment: Employing a small team of developers or outsourcing the development can result in a substantial cost. According to Glassdoor, the average annual salary for an AI developer in the US is approximately $114,000. For small projects, the development timeline may span a few months, impacting the overall cost.

Maintenance and Updates: Continuous maintenance and updates are essential to keep the AI system operational and relevant. This could add an additional 10-20% to the initial development cost annually.

AI Software Costing for Medium-Sized Enterprises

Medium-sized enterprises generally have more resources and a broader scope for implementing AI systems. The cost for such businesses can vary from $50,000 to $500,000. Here is a breakdown of the factors influencing these costs:

Advanced AI Solutions: Medium-sized enterprises often require more advanced AI solutions such as machine learning models for customer insights, fraud detection systems, or advanced automation tools.

Data Management: The volume of data to be managed is larger, necessitating more robust data management systems. This includes expenses for data warehousing, data processing, and ensuring data security.

Development Team: Building an in-house team of AI experts, data scientists, and engineers can be costly. According to Indeed, the average annual salary for a data scientist in the US is around $122,000. The size of the team and the duration of the project will impact the total cost.

Infrastructure: Investment in high-performance computing infrastructure, cloud services, and software licenses is necessary. Cloud platforms like AWS, Google Cloud, or Azure offer AI services that can cost between $0.10 to $3 per hour, depending on the service.

AI Development Cost Breakdown
- Custom or Off-the-Shelf – $5000-$300,000
- Prototype Development – Starts from $25000
- Software Cost – $30,000-$50,000
- Maintenance – Upwards of $60,000/year
AI Development Costing For Large Corporations

For large corporations, the cost of building an AI system can surpass $1 million. The complexity and scale of AI solutions for these businesses require significant investment. Here are some factors contributing to these costs:
- Complex AI Solutions: Large corporations may implement AI for various purposes such as supply chain optimization, customer service automation, predictive maintenance, and more. These systems require extensive development and testing.
- Big Data Handling: Managing and processing vast amounts of data is crucial. This involves significant investment in big data technologies and infrastructure.
- Expert Team: Hiring top-tier AI experts, including PhD-level researchers and experienced engineers, is expensive. According to ZipRecruiter, AI researchers can earn up to $165,000 annually.
Integration with Existing Systems: Integrating AI systems with existing IT infrastructure can be complex and costly. This includes software development, testing, and ensuring seamless operation with other enterprise systems.

Compliance and Security: Ensuring that AI systems comply with industry regulations and are secure from cyber threats adds to the cost. This involves regular audits, security upgrades, and compliance checks.

Factors Influencing AI System Costs

Several factors influence the cost of building an AI system, regardless of business size:
- Scope and Objectives: The broader the scope and the more ambitious the objectives, the higher the cost.
- Technology Stack: The choice of technology stack, including programming languages, frameworks, and tools, impacts the cost.
- Custom vs. Off-the-Shelf Solutions: Custom AI solutions are more expensive but tailored to specific business needs, whereas off-the-shelf solutions are cheaper but less flexible.
- Development Timeline: Longer development timelines can increase costs due to prolonged resource utilization.
- Post-Deployment Costs: These include maintenance, updates, scaling, and user training.
Conclusion

In conclusion, the cost of building an AI system varies significantly based on the type and size of the business. Small businesses might invest between $10,000 and $50,000, medium-sized enterprises between $50,000 and $500,000, and large corporations over $1 million.

The factors affecting these costs include the type of AI solution, data management, development team, infrastructure, and ongoing maintenance. According to my research, investing in AI can bring substantial benefits, but it is crucial to plan and budget appropriately to ensure successful implementation. For more detailed insights, you can refer to resources such as Forbes, Gartner, and McKinsey.

Did you know that the AI market is projected to reach nearly 2 trillion USD by 2030? This growth is not surprising given the rapid expansion and transformation of industries by AI.

Have you ever thought about the expenses associated with AI development?

Understanding the cost of AI development is essential for businesses and individuals looking to utilize this powerful technology. It can aid in resource allocation, budgeting, and evaluating the feasibility and return on investment of AI initiatives.

In this article, you will discover various factors that impact the cost of AI. Keep reading to make well-informed decisions.

What is AI?

Artificial Intelligence involves creating intelligent systems capable of performing tasks that typically require human intelligence. These systems use advanced algorithms and techniques to analyze data and solve complex problems. AI encompasses various technologies such as machine learning, natural language processing, and more.

Main Components of Artificial Intelligence.

Source

Factors Influencing AI Development Costs

Below are specific factors that influence the cost of AI development:

1. Type of AI:

The type of AI solution being developed significantly affects the cost. More advanced AI models generally require additional resources and expertise, leading to increased costs. Here are some common types of AI and their impact on pricing:

Rule-Based Systems: These systems follow predefined rules and logic to make decisions or perform tasks. They are relatively simpler and less expensive to develop compared to other AI types. They require a well-defined set of rules and guidelines, which can be established with less effort and resources.

Machine Learning Models: Training AI models on data to learn patterns and make predictions or decisions is involved in machine learning. Developing machine learning models requires expertise in data analysis and model training. The cost can vary based on factors such as model complexity, data volume, and the need for specialized algorithms.

Deep Learning Networks: Deep learning is a subset of machine learning that uses artificial neural networks with multiple layers to process complex data. Deep learning models are highly sophisticated, requiring significant computational power and extensive training data. Developing deep learning networks can be more expensive due to the need for advanced hardware and specialized expertise.

Natural Language Processing (NLP): NLP focuses on enabling computers to understand and process human language. Developing NLP systems involves language parsing, sentiment analysis, and generation. The cost depends on the complexity of language processing requirements and the desired accuracy level.

2. Solution Complexity:

The complexity refers to the training data and processing power required to solve a problem. Assessing the complexity upfront can help in setting realistic expectations and budgets for the development process.

Here are some factors that can impact the complexity of AI development:

Algorithm Complexity: Developing AI systems with complex algorithms, such as those used in deep learning or advanced machine learning models, necessitates specialized expertise. These algorithms may involve intricate mathematical computations and complex optimization techniques. Implementing such algorithms adds complexity and significantly impacts AI development costs.

Integration with Multiple Systems: Integrating AI systems with existing software applications requires seamless communication and data exchange between components. The involvement of a higher number of systems or applications increases the complexity and development cost.

Real-Time Processing or Decision-Making: Some AI solutions must process and analyze data in real-time to make instant decisions or provide real-time responses. Implementing real-time capabilities adds complexity to the system architecture, potentially requiring additional resources, infrastructure, and expertise, thereby affecting the cost.

User Interface and User Experience: If the AI solution requires a user interface or user experience design, the complexity of designing an intuitive and user-friendly interface can impact the development cost. Creating visually appealing and interactive interfaces with smooth user interactions may require additional time and resources.

3. Data Volume:

AI systems depend on large volumes of data to learn and enhance their performance. Acquiring, cleaning, and organizing the necessary data can involve significant costs, especially when the data is scarce or needs to be collected from various sources.

Here are some references related to the amount of data:

Data Quantity: AI systems require substantial data for training and learning. However, obtaining large volumes of data can be costly, especially if the data needs to be acquired from external sources or requires extensive data collection efforts.

Data Quality: The quality of data used for developing AI is critical. High-quality data that accurately represents the problem domain leads to improved AI performance. Ensuring data quality may involve tasks such as data cleaning, preprocessing, and validation, which can increase development costs.

Data Diversity: Having diverse data covering a wide range of scenarios and variations can enhance an AI system’s ability to handle different situations. However, collecting or curating various datasets may result in additional costs, especially if the desired data is not readily available.

Data Accessibility: The ease of accessing required data can impact development costs. If the data is readily available in a well-organized format, the cost of acquiring and processing it may be lower. However, if the data is scattered across various sources or needs to be extracted from different formats, it will require extra effort, thus adding to costs.

Data Privacy and Security: Ensuring data privacy and security is crucial when working with sensitive or personal data. Implementing appropriate measures to protect data privacy can increase development expenditure.

Expert Services: AI development often requires specialized expertise. While expert services may increase costs, they provide valuable knowledge and skills that can significantly impact the success of the AI project.

AI Professionals: Skilled AI professionals possess the knowledge and expertise to develop AI systems. Hiring experienced AI professionals can increase development costs as their expertise comes at a premium. Their skills in algorithm development, data analysis, model training, and system optimization contribute to the overall quality and performance of the AI solution.

AI Development Companies: Partnering with AI development companies can provide access to a team of experts specializing in AI development. These companies have experience developing AI solutions across various industries and can offer valuable insights and guidance throughout the project. Moreover, they have extensive knowledge of optimization techniques and can fine-tune the AI system.

Quality Assurance and Testing: Ensuring the quality and reliability of AI systems is crucial. Expert services for quality assurance and testing can help identify and resolve issues. They can also validate results and ensure the system meets the desired objectives. These services contribute to the overall cost but help deliver a robust and reliable AI solution.

Training and Maintenance: Training and Maintenance are essential aspects of AI development that require ongoing effort and investment. Ignoring training and maintenance can lead to decreased efficiency or even system failure.

Regular Updates: AI models must be regularly updated to incorporate new data, algorithms, or features. Updating the model helps improve its performance and adaptability to changing conditions. Updating the AI system may require additional development time and resources, contributing to the overall cost.

Monitoring and Performance Evaluation: Continuous monitoring of the AI system’s performance is necessary to identify any issues or deviations. Regular evaluation helps ensure the system functions optimally and meets the desired objectives. Monitoring and evaluation activities may involve data analysis, performance metrics assessment, and fine-tuning, all of which incur costs.

Troubleshooting and Bug Fixing: Like any software system, AI solutions may encounter issues or bugs that must be addressed. Troubleshooting and bug fixing involve identifying and resolving system malfunctions or errors. These activities require skilled professionals and may involve minor or significant costs depending on the complexity of the problem.

Data Management: Managing and updating the data for AI training is required to maintain the system’s accuracy and relevance. This includes data collection, cleaning, labeling, and organizing. Data management activities can contribute to the ongoing cost of maintaining the AI system.

Costs Associated with AI: Implementing AI involves various expenses that need to be considered, some of which are as follows:

1. Hardware Costs: Hardware costs in AI development refer to the expenses associated with the physical infrastructure required to support AI systems. These costs can include:
- High-Performance Computing Devices
- Specialized Hardware Accelerators
- Storage Solutions
- Networking Infrastructure
- Cloud Computing Services
2. Software Costs: Software costs are the expenses associated with acquiring, using, and maintaining software systems. These costs can include:
- Licensing Fees for AI Development Tools
- Subscriptions for AI Frameworks
- Software Maintenance and Support Costs
- Customized Software Development Expenses
- Integration Costs for Software Components
- Charges for Software Upgrades and Updates
Labor expenses are linked to the workforce involved in a project or operation, which can stem from hiring specialized AI professionals, paying salaries or consulting fees, training existing staff or hiring additional team members, conducting research and development activities, allocating resources for project management and coordination, as well as ongoing collaboration and communication among team members.

Training and maintenance are ongoing processes for AI systems, and the costs incurred for these activities include data labeling expenses, computational resource costs, monitoring and optimization fees, as well as software updates and upgrades.

In addition to the core development and maintenance expenses, there may be additional costs associated with AI development, such as data acquisition and cleaning costs, integration with existing systems, infrastructure setup, and necessary security measures.

The cost of developing artificial intelligence can vary significantly based on the technology being developed or implemented, the scope and complexity of the project, the level of expertise required, and the specific industry or application. These costs can range from as low as $900 to well over $300,000, but these figures are only general estimates.

Here’s a breakdown of the primary cost considerations for AI under relevant subheadings:

Research and Development (R&D) involves significant research and experiments, requiring a dedicated team of experts, including salaries, equipment, software, and data acquisition.

AI algorithms rely on large amounts of high-quality data for training, and preparing and curating the data can involve costs related to data collection, cleaning, labeling, and storage.

Building and fine-tuning AI algorithms may require specialized expertise, including data scientists, machine learning engineers, and software developers, with costs depending on the complexity of the algorithms and the time required for development.

AI models may require powerful computational resources, such as GPUs (Graphics Processing Units) or specialized AI chips, to process and analyze data efficiently, leading to significant costs for acquiring and maintaining these hardware components.

Many organizations utilize cloud computing platforms to leverage their AI capabilities, and the costs can vary depending on usage, storage, and processing requirements.

Deploying AI systems within existing infrastructure may involve integrating with existing software, databases, or APIs, the cost of which depends on the complexity and compatibility of the integration process.

AI models often require training on specific datasets to optimize performance, with costs related to the time and resources required to train the models, as well as the testing and validation processes.

Tailoring AI solutions to specific business needs or industries may involve additional development and configuration costs.

AI systems require ongoing maintenance, updates, and monitoring to ensure optimal performance and security, including costs related to bug fixing, algorithm improvements, and infrastructure maintenance.

Providing training and support for end-users or employees who interact with AI systems may require additional resources and associated costs.

Organizations must ensure AI systems comply with ethical guidelines and legal requirements, which may involve costs related to data privacy, bias mitigation, and transparency measures.

The cost of AI can vary significantly depending on the specific project and context, with some AI solutions readily available as pre-built services or open-source frameworks, reducing development costs. Additionally, as AI technologies advance and become more widespread, the overall cost of implementation and deployment may decrease over time.

It’s important to thoroughly analyze the requirements, project scope, and desired outcomes to estimate the precise cost of developing AI.

To unlock the immense potential of AI, it’s crucial to invest in the future today with the support of an Adaptive AI development company like Parangat Technologies, an esteemed Enterprise AI Development Company. Embracing AI technologies can empower businesses to achieve unparalleled efficiency, data-driven decision-making, and enhanced customer experiences.

“By leveraging the knowledge and skills of firms such as Parangat Technologies, businesses can take advantage of the revolutionary potential of AI, guaranteeing that they stay competitive and forward-thinking in a constantly changing environment. AI represents the future of both business and technology, and the present is the opportunity to enjoy time to invest in it and its advantages.”
November 7, 2024
How did ChatGPT attain widespread popularity so rapidly?

China’s tech giant Alibaba wants to get involved in the artificial intelligence business. At the same time, Beijing is preparing state regulations. But governments in the West must also ask themselves: How many regulations does the technology need?

It was a big announcement for the Chinese internet giant Alibaba. The cloud division of the online retail group today presented a competitor to the text robot ChatGPT: the voice software “TongyiQianwen”, which means something like “truth from a thousand questions”, which also uses artificial intelligence (AI). But shortly afterwards, the developers’ joy was probably dampened. At the same time, the Chinese internet regulator, the “Cyberspace Administration of China”, published the first draft of planned regulations for AI services.

In 21 points, the authority presents possible requirements that could soon be imposed on Chinese companies and developers of AI language models. According to Beijing’s wishes, the content must reflect the “basic values of socialism”. In addition, no information may be disseminated that could disrupt the economic and social order. When developing the algorithms, care should also be taken to prevent discrimination based on gender or age, for example.

Bot with “hallucinations”

One problem for developers is the rule that all content must be truthful. The development of AI language models is still at an early stage.In many cases, the software is still imprecise and prone to errors. Google made an embarrassing mistake when introducing its chatbot “Bard”,which gave an incorrect answer about the James Webb telescope in its first public appearance . Alibaba’s chatbot, on the other hand, is initially geared towards business life and is intended to write documents or emails, for example.

However, it remains to be seen how well the bot will fare in the race against the competition, says George Karapetyan, AI expert at the consultancy LPA, to tagesschau.de . “According to initial user reports, Alibaba’s bot has also already had ‘ hallucinations’, which ultimately means that it confidently gives incorrect answers.”

The Chinese regulator now wants to put a stop to such false content. Comments and suggestions on the catalog of regulations can be submitted until May 10. “As the Chinese government begins to regulate and dictate what these bots can and cannot say, this could represent an additional hurdle in balancing innovation with compliance,” said Karapetyan.

Is developing technology too quickly?

From the expert’s point of view, the early introduction of clear rules for companies can also be helpful in reducing the risk of unforeseen results. “If China succeeds in defining clear guardrails early on, this also presents opportunities.” However, it can be difficult to regulate a technology that is developing so quickly and is so intelligent. Every day there are reports of how Internet users are circumventing theprotective mechanisms for controlling bots.

Alibaba is just the latest example of a Chinese company with its own text robot. Just one day earlier, the Hong Kong-based AI company SenseTime presented its chatbot “SenseChat” in a live demo, to which the stock market reacted with a strong increase in share prices. And last but not least, the Chinese search engine Baidu also demonstrated its chatbot”Ernie Bot”, which, however, generated less enthusiasm and a falling share price.

“Chinese bots are currently lagging behind and are primarily focused on the Chinese language,” says AI expert Karapetyan. At the moment, ChatGPT, the software designed by the start-up OpenAI and supported by Microsoft, is the “clear market leader and the gold standard”among chatbots.

The rapid advances of artificial intelligence are causing both excitement and apprehension. An intriguing interview conducted by CBS News with Google’s AI executives examines both perspectives.

Artificial intelligence (AI) is progressing rapidly. One striking example of the impressive – and in some ways unsettling – advancement is Google Bard. This AI-based chatbot was created by Google in response to the success of OpenAI’s ChatGPT and was released in a limited capacity in March 2023.

Bard swiftly generates a rich human-like narrative with its own characters in response to a six-word prompt – all within seconds. Over several months, the AI has extensively studied the content available on the Internet, forming a model of language. Instead of searching, responses are derived from this language model, thanks to Bard’s microchips, which operate at a speed 100,000 times faster than the human brain.

On one hand, there is excitement regarding the current capabilities of AI and the anticipation of how it will further simplify our professional lives in the future. Conversely, there are concerns about the rapidly evolving professional landscape and the potential for AI to surpass humans, potentially causing more harm than good (key term: machine learning, ML).

The most significant transformations are expected to occur in work environments. According to James Manyika, senior vice president of Google, over two-thirds of individuals will likely witness changes in their job descriptions. These jobs won’t vanish due to the integration of AI and automation but will undergo transformation. We are on the brink of significant changes that will impact skill sets, requiring individuals to adapt to working alongside machines.

One of the key concerns in the continued progression of AI is likely how to develop AI systems driven by human values. Sundar Pichai, CEO of Google LLC and its parent company Alphabet Inc., has emphasized the involvement of not only engineers but also social scientists, ethicists, philosophers, and others in the development process.

However, he also noted that the societal decision-making process should unfold during the development of AI and should not rest solely on the choices made by any one company.

ChatGPT, a human-like AI chatbot, has gained widespread attention across social media in recent days. OpenAI’s ChatGPT has rapidly gained popularity, sparking widespread discussions across the internet. It is built on artificial intelligence and possesses the ability to respond to queries, engage in natural conversations, and much more.

In just five days, it has garnered millions of users. Developed by the AI research company OpenAI, this chat tool, supported by Microsoft and Elon Musk, utilizes the company’s GPT3 (Generative Pre-Trained Transformer 3) technology, enabling users to converse with the AI on a wide range of topics.

It stands out from previous AI chat tools due to its ability to deliver responses in natural-sounding language – to the extent that if one wasn’t aware, they could easily mistake it for a conversation with a real human being.

Individuals have showcased how the AI assists them in tasks beyond basic conversations, such as composing articles and academic papers, drafting complete job applications, and even aiding in coding.

At present, it is available for free trial upon registration using an email and phone number. However, OpenAI mentions that conversations are reviewed “to enhance our systems” and may be used to train AI.

How did ChatGPT attain widespread popularity so rapidly?

According to Adam Conner, vice president for technology Policy at the Center for American Progress, ChatGPT quickly gained popularity because it was among the first AI technologies of its kind to be publicly accessible in a manner understandable to the general public.

“What sets GPT apart is its generative nature – it produces outputs in a manner comprehensible to ordinary individuals as opposed to simply outputting code or data,” Conner clarified.

Unlike traditional search engines like Google, ChatGPT can engage in conversation, offering human-like responses and dialogue with users. Users can request ChatGPT to generate a resignation letter, prompts for class discussions, and even academic tests.

ChatGPT can be likened to a “virtual companion,” as described by Jim Chilton, CTO of Cengage Group, an education technology company.

“I replicated a similar action with a calculus example, ‘generate a calculus final exam for me.’ It not only created the exam but also provided solutions to all the problems. It systematically explained the steps for solving the calculus problems, reinforcing the principles throughout the process.”

While some advocate for a temporary or justified ban due to the widespread use of ChatGPT among students, experts and educators argue that bans are not effective or equitable in the long run.

Though Conner recognizes the purpose of bans on ChatGPT, he adds that “everyone acknowledges that it’s not a universal solution.”

Glantz highlighted one significant issue with bans, which is “equity and access.”

How do governments respond?

Microsoft and ChatGPT’s competitors in the tech industry are under pressure to push ahead with their artificial intelligence business, even if the product is still immature. At the same time, given the rapid development, pressure is growing on governments around the world to find answers to the question of how lawmakers should respond.

In the USA, the IT authority NTIA (“NationalTelecommunications and Information Administration”) today announced publicconsultations on possible government measures. “Just as food and cars only come onto the market if their safety is guaranteed, AI systems should also give the public, the government and companies the assurance that they are fit for purpose,” it said in a statement. The authority could ultimately recommend safety assessments or certification of artificial intelligence to politicians.

Italy sets a deadline for ChatGPT

The EU is also looking for government regulations for the new technology. Most recently, the Italian data protection authority caused a stir by temporarily blocking ChatGPT in the country . The main concerns were the massive collection of personal data and the protection of minors. Italy has given OpenAI 20 days to inform the company of its further measures. Otherwise, it could face a fine of up to 20 million euros or four percent of annual turnover.

Two years ago, the EU Commission presented a draft AI regulation that could come into force this year. Regulation is urgently needed in this area, says Paul Lukowicz, head of the Embedded Intelligence research area at the German Research Center for Artificial Intelligence (DFKI) tagesschau.de .The technology will change the world in ways that we cannot even imagine today.Therefore, we cannot simply let it run its course in the sense of”uncontrolled growth”.

When a school bans ChatGPT, it can only be utilized on school computers and WiFi. Although this benefits students without access to technology outside of school, many students have personal devices at home through which they can use AI technology. According to Glantz, when a program like ChatGPT is prohibited on school computers and WiFi, it impacts students who solely rely on school technology for accessing technology when they are at school. Glantz asserts that some students have resorted to using a school WiFi hotspot to bypass the ban.

It is also essential to teach students how to utilize ChatGPT as this kind of technology might be necessary for future employment. Glantz stated, “ensuring that we equip the students with the necessary skills to leverage technology will be crucial.”

The maneuvering around or with ChatGPT could be the initial step in defining the relationship between schools and AI technology.

Conner suggests that decisions regarding the incorporation of ChatGPT and AI in schools in the future will need to involve the company, educators, parents, and administrators to be made.

ChatGPT, the AI chatbot, swiftly gained immense popularity in just a few weeks—much faster than social media platforms such as TikTok or Instagram. Only two months after its late November launch, the chatbot had 100 million monthly active users by January, as per Similarweb’s data. A study by Swiss bank UBS pointed out that “in 20 years within the internet space, we cannot recall a faster ramp in a consumer internet app.” According to Digital-adoption.com, OpenAI, the owner and host of ChatGPT, recently joined the list of the 50 most visited websites globally.

To provide context, Instagram took two and a half years to reach 100 million, while TikTok achieved this milestone in nine months.

The rapid rise of ChatGPT underscores its utility in assisting with various tasks and the widespread curiosity about human-like machines. Experts are divided on whether this signifies the beginning of a new AI era or if the excitement will diminish as people reach the limits of ChatGPT’s current capabilities.

Here’s why ChatGPT gained widespread popularity quickly and what that implies for the future.

What is ChatGPT?

ChatGPT, a chatbot developed by the San Francisco company OpenAI, is categorized as a generative AI. It swiftly and clearly responds to almost any prompt. Unlike many chatbots that only know how to respond to specific keywords or triggers, ChatGPT can provide comprehensive, essay-length answers on virtually any topic.

ChatGPT accomplishes this by processing the vast amount of data on the Internet through powerful neural networks, which are software loosely modeled on the neurons in the human brain. While this technology has been in existence for several years, Yann LeCun, the chief AI scientist at Meta, recently argued that ChatGPT was “not particularly innovative” and largely relied on Google’s Transformer neural net technology unveiled in 2017.

Some experts are surprised about the explosive popularity of ChatGPT. Margaret Mitchell, the chief ethics scientist at the AI company Hugging Face, stated that “the technology wasn’t introducing any fundamental breakthroughs.” However, ChatGPT was the first major project to introduce such AI for public use, experimentation, and testing. Unlike other companies like Google, which held back due to the unpredictability of this new technology and the potential harms it could cause, such as the spread of misinformation or hate speech, OpenAI chose to hurriedly bring their product to the market this fall in the face of potential upcoming competition, as reported by the New York Times.

While ChatGPT is built on complex technology, its visual interface is highly user-friendly: users simply enter text into a text box, similar to using Google. This straightforward interface has enabled people of all ages and backgrounds to immediately interact with it. Another strength of ChatGPT is its adaptability. If a user is dissatisfied with its response to their prompt, they can modify their input, and the AI will adjust accordingly.

What are people doing with ChatGPT?

The initial reason for ChatGPT’s viral spread was its novelty. Users requested ChatGPT to create a biblical verse about removing a peanut butter sandwich from a VCR or to come up with fantasy weapons inspired by Elvis. In just seconds, the AI would generate options such as “Love Me Tender Dagger” and “Blue Suede Sword.”

However, ChatGPT’s use quickly expanded beyond memes and tricks, extending into professional applications. ChatGPT is capable of brainstorming ideas, writing articles, and coding. People began using it to compose entire job applications, curriculums, academic papers, and scripts in various programming languages. According to Similarweb’s data, programming and developer software have emerged as some of the main uses for ChatGPT.

According to TIME, Sean Ellul, one of the co-founders of Metaverse Architects, mentioned in an email that ChatGPT has significantly improved their productivity and creativity, and he uses it for various tasks such as brainstorming, coding, writing articles, and generating new project ideas. The technology has prompted several companies, including Buzzfeed, to modify their business models to incorporate it into their workflows, particularly for quizzes and personalized content.

As a result of concerns about AI-generated school assignments, school districts across the United States, including New York City, have banned the use of ChatGPT.

Due to a substantial surge in interest, OpenAI has been forced to reject numerous users, redirecting them to a message stating, “ChatGPT is at capacity right now.” A paid tier has been introduced to address this issue, providing access to users during peak periods.

Could this be just the beginning of the widespread adoption of generative AI technology?

Following the surge in interest in ChatGPT, competitors in the technology sector are hastily introducing their own versions. Google has responded to ChatGPT by announcing its own Bard AI, which is set to launch in the upcoming weeks. Similarly, the Chinese tech giant Baidu is preparing to release a comparable chatbot in March, and Anthropic, an AI company founded by former OpenAI employees, has secured hundreds of millions in funding.

Microsoft, an investor in OpenAI, is in the process of integrating ChatGPT into its Bing search engine and Teams messaging platform. Consequently, many everyday work processes are likely to be augmented by generative AI technology, often without users’ awareness.

However, there are potential risks on the horizon. AI has been involved in generating hate speech, spreading misinformation, and assisting in the creation of malicious code. According to Mitchell, as the initial excitement surrounding this technology wanes, criticisms of its problematic applications are likely to increase.

Mitchell is apprehensive about the potential impact of ChatGPT on individuals seeking mental health guidance. She believes that ChatGPT might offer toxic or bullying advice without understanding the consequences, as it lacks comprehensive knowledge of the world.

Furthermore, she is worried about its usage as a substitute for search engines, as ChatGPT may provide declarative but false information. It has even fabricated a detailed history of a “successful civilization” created by dinosaurs. Mitchell is concerned that people are more likely to accept automated responses as factual due to cognitive bias.

The current AI arms race sparked by ChatGPT’s rapid rise could lead its competitors to take shortcuts in order to gain market share. Mitchell is concerned about the potential consequences, as she believes that regulatory measures are often reactive and tend to follow significant negative events.

When asked whether artificial intelligence is developing too rapidly, a chatbot may avoid giving a direct response, whereas high-profile tech leaders and researchers may firmly assert that it is indeed growing too fast.

According to Bard, Google’s AI engine, there is no straightforward answer to this question due to its complex nature and diverse perspectives.

Nevertheless, prominent figures in the tech industry have expressed the need to slow down the development of AI. This could involve companies establishing standards and disclosing their current and future use of AI, as suggested by business leaders.

In a letter signed by over 1,800 individuals, including Elon Musk, the CEO of Tesla and Twitter, and Steve Wozniak, a co-founder of Apple, as well as researchers from renowned universities like Harvard and Oxford, the rapid adoption of AI without fully understanding its implications was highlighted as a major concern.

“In recent months, there has been a race among AI labs to create and deploy increasingly powerful digital minds that are difficult for anyone, including their creators, to understand, predict, or control,” states the letter.

The letter acknowledges the need for engineers to develop AI systems, but the concern is the absence of agreed-upon guidelines for the operation of models such as ChatGPT, GPT4, Bard, and other generative AI systems.

It urges the development of powerful AI systems only when there is confidence in their positive effects and manageable risks.

To achieve this, companies like SAP, the German software giant that assists businesses with financial reporting, inventory tracking, and human resources services, are establishing standards for their teams. Others like PwC, the global accounting and consulting firm, advise CEOs to be transparent about their integration of the technology.

Sebastian Wieczorek, vice president of artificial intelligence technology and global lead of AI ethics at SAP, stated, “AI is a rapidly evolving technology that presents new opportunities every day.”

“All businesses should ask themselves if they understand the actions of AI,” commented Wes Bricker, a vice chair at PwC.

“AI will revolutionize major aspects of business,” he added, while emphasizing the responsibility of business leaders to be transparent as they gain more knowledge about AI.

The fast-paced nature of AI and its unforeseen consequences are well known. Consider Bing’s Sydney AI chatbot or Goldman Sachs’ announcement that AI could potentially boost annual world GDP by 7%.

Wieczorek described SAP’s approach as an ongoing evolution, emphasizing continuous improvement and the steps taken to utilize available data. “What benefits can we achieve?” “What is the accuracy we can attain with current technologies?” These are the questions SAP teams are addressing.

Bricker stressed the need for business leaders to enhance the regulations governing AI systems and processes. “Do we have clear governance guidelines to understand and prevent misuse or overuse?” he inquired, emphasizing the importance of AI being “understandable and explainable.”

AI extensively utilizes sensitive data, and according to Bricker, businesses have a duty to safeguard this data. He further added that it is vital to understand how AI might impact experience or security.

Businesses and consumers have various reasons to be enthusiastic about and embrace AI. Wieczorek mentioned that AI could help address common business challenges related to internal and external communications, finance, HR processes, promotions, training, and retirement planning.

SAP focuses its AI development on improving and standardizing everyday business processes. Wieczorek highlighted the necessity for engineers to train the programs on different types of data, such as images, and noted that these models, although seemingly basic, are currently limited in comparison to human capabilities.

According to Wieczorek, any AI ethics policy should prioritize human support in decision-making. For every use case, SAP requires a series of risk assessment questions, particularly relating to the processing of personal and sensitive data.

Bard also reflects on the potential impact of AI. “I recognize that AI has the potential to pose risks, but I am optimistic about its potential for good and believe that it can be developed in a way that minimizes risks and maximizes benefits.”

Artificial Intelligence (AI) has evolved from a theoretical concept to a disruptive force that is transforming industries globally. Recent years have seen a rapid acceleration in AI development, leading to discussions and speculation about the reasons behind this progress.

Having dedicated considerable time and effort to understanding the complexities of AI through programs such as INSEAD and various others, I have observed the impressive speed at which AI has advanced.

In this piece, we will analyze the primary factors propelling the acceleration of AI, offering valuable insights into this transformative phenomenon.

1. Technological Progress:

– The growth in computing power, driven by Moore’s Law and advancements in hardware architecture, has unlocked unprecedented capabilities for AI systems. For example, NVIDIA’s latest A100 GPU provides up to 20 times the performance of its predecessor, the K80 GPU, in deep learning activities.

– Specialized AI accelerators like Google’s Tensor Processing Units (TPUs) and Intel’s Nervana Neural Network Processors (NNPs) have further expedited AI computations, delivering performance gains surpassing traditional CPU architectures by significant margins.

– Innovations in algorithms, particularly in deep learning, have transformed AI applications in various domains, such as natural language processing and image recognition. For instance, breakthroughs like the development of Transformer models like BERT and GPT (Generative Pre-trained Transformer) have markedly enhanced AI’s ability to comprehend and generate human-like text.

– Advancements in Natural Language Processing (NLP), including the introduction of pre-trained language models like OpenAI’s GPT series and Google’s BERT, have led to substantial performance enhancements in NLP tasks, making state-of-the-art capabilities more accessible.

Ref.: NVIDIA’s annual GPU Technology Conference (GTC) presentations, OpenAI’s research publications, academic papers from conferences like NeurIPS and ICML.

2. Abundance and Quality of Data:

– The widespread use of digital devices and IoT sensors has generated vast volumes of data, which serve as the lifeblood of AI algorithms. It is estimated that by 2025, the global datasphere will expand to 175 zettabytes, presenting significant opportunities for AI applications.

– Improved data collection methods and data cleaning techniques have raised the quality and relevance of datasets, facilitating the development of more accurate AI models. According to a McKinsey report, organizations that utilize data-driven insights are 23 times more likely to acquire customers and six times more likely to retain them.

– The adoption of cloud computing has further accelerated the data abundance trend by providing scalable storage and computing resources. For instance, Amazon Web Services (AWS) offers services like Amazon S3 for storage and Amazon EC2 for computing, enabling organizations to store and process large datasets in the cloud with flexibility. This scalability and flexibility empower businesses to handle fluctuating data volumes and conduct complex AI analyses without substantial upfront investments in infrastructure.

3. Economic Conditions:

– During times of economic weakness, businesses might seek to enhance efficiency and productivity through the adoption of AI. According to reports from major firms such as Gartner, Forrester, and McKinsey, AI technologies present opportunities for optimizing resources and mitigating risks, which could be especially valuable in times of economic decline.

– The anticipation of tangible financial returns on AI investments is a significant driving force behind the increase in AI adoption. Businesses are increasingly realizing the potential of AI technologies in driving revenue growth, reducing costs, and gaining competitive advantages. Investments in AI are motivated by the expectation of concrete benefits, including enhanced operational efficiency, improved customer experiences, and better decision-making capabilities.

4. Government and Public Investment:

– In order to foster economic growth and competitiveness, governments worldwide are progressively investing in AI research and development. For example, China’s “New Generation Artificial Intelligence Development Plan” strives to lead global AI innovation by 2030, particularly in strategic sectors such as healthcare, transportation, and defense.

– Through public-private partnerships such as Canada’s Pan-Canadian AI Strategy and the U.S. National Artificial Intelligence Initiative, significant resources are dedicated to AI research, talent development, and infrastructure, promoting collaboration between academia, industry, and government agencies.

– Singapore has taken the lead in AI investment and innovation, committing more than $500 million to its national AI strategy. Initiatives like AI.SG, a program initiated by the government, unite stakeholders from academia, industry, and government agencies to advance AI research, talent development, and adoption across various sectors.

5. AI Platforms and Innovations:

– The growth of AI platforms has been significant in recent years. Reports from the industry indicate that there are now over 500 AI platforms available, a marked increase from just 100 platforms two years ago. These platforms, such as Sora, Dall-e, and Claude, offer advanced AI capabilities like natural language processing, computer vision, and generative modeling, catering to a wide range of use cases and industries.

6. Current Crisis and Wars:

– AI technologies are being utilized in current conflicts worldwide for surveillance and targeting purposes. For instance, AI-enabled drones are being used for reconnaissance and targeted strikes, while social media platforms leverage AI algorithms to manipulate public opinion and spread misinformation.

– In the realm of cybersecurity and cyber warfare, state and non-state actors are increasingly employing AI-powered tools for offensive and defensive purposes, conducting activities such as espionage, sabotage, and cyber attacks. Autonomous malware, AI-driven phishing attacks, and adversarial machine learning techniques pose significant threats to national security and critical infrastructure.

BUT….Limitations of AI and the Importance of Responsible AI:

– While AI offers great potential, it also presents limitations and ethical considerations. AI systems can demonstrate biases, lack transparency, and be vulnerable to adversarial attacks. Moreover, the deployment of AI in critical domains like healthcare and criminal justice raises concerns regarding privacy, fairness, and accountability.

– The development of responsible AI involves addressing these challenges through robust ethical frameworks, transparent algorithms, and inclusive decision-making processes. Initiatives like the AI Ethics Guidelines by the European Commission and the Responsible AI Institute are aimed at promoting ethical AI development and deployment practices.

Conclusion:

The recent rapid advancement of AI represents a convergence of technological, economic, and societal factors, pushing us into an era of unparalleled innovation and disruption. As AI continues to progress and infiltrate every facet of our lives, it is crucial to remain mindful of its implications and effects. While the potential benefits of AI are immense, including exhaustive productivity, efficiency, and economic growth, we must also address its limitations and ethical considerations.

Realizing the full potential of AI demands a collaborative effort from stakeholders across industries, academia, governments, and civil society. Through cultivating a culture of responsible AI development and deployment, we can mitigate risks, ensure fairness and accountability, and maximize the societal benefits of AI technologies.

In summary, the acceleration of AI is not solely a technological advancement but a societal transformation that requires thoughtful consideration and strategic action. By harnessing the driving forces behind the surge in AI while upholding ethical principles and inclusivity, we can pave the way toward a future where AI serves as a powerful tool for positive change and human progress.

The Chinese technology firm Alibaba launched over 100 new open-source artificial intelligence models and text-to-video AI technology on Thursday, ramping up its efforts to compete in the rapidly growing field of generative AI. The new open-source models come from Alibaba’s Qwen 2.5 family, which is the company’s latest foundational large language model that was released in May.

Similar to their U.S. counterparts, Chinese tech companies are heavily investing in generative AI, with businesses racing to create strong product portfolios and diversified offerings. While rivals like Baidu and OpenAI have largely taken closed-source approaches, Alibaba has adopted a hybrid strategy, investing in both proprietary and open-source developments to expand its AI product range.

These new models vary in size, ranging from 0.5 to 72 billion parameters, which affect an AI model’s capabilities and performance, and they offer proficiency in mathematics, coding, and support for over 29 languages, according to a statement from Alibaba.

The models are designed to serve a wide variety of AI applications across different sectors, including automotive, gaming, and scientific research. On Thursday, Alibaba also introduced a new text-to-video model as part of its Tongyi Wanxiang image generation family, entering a market that an increasing number of Chinese tech firms are exploring. This move places Alibaba in direct competition with global entities like OpenAI, which is also interested in text-to-video technology.

During J.P. Morgan’s 20th annual Global China Summit in May, Alibaba Group Chairman Joe Tsai emphasized the value and potential unlocked by artificial intelligence.

At the conference in Shanghai, over 2,700 delegates from 1,300 companies across 33 markets gathered to gain insights from sectors like tech, healthcare, and renewables.

In Tsai’s fireside chat, AI was a prominent topic of conversation.

“AI is an extremely important field where you can’t just choose one path,” noted Tsai, who spoke next to Kam Shing Kwang, Chairwoman for North Asia and Vice Chair of Investment Banking for Greater China at J.P. Morgan.

“We are the only company [in China] that operates a leading cloud business while remaining competitive in AI,” he remarked. “The combination of AI and cloud services is crucial.”

During a 30-minute dialogue with Kwang, Tsai elaborated on how AI is propelling growth in the company he co-founded 25 years ago, influencing both Alibaba’s core e-commerce operations and its cloud services.

“We see immense potential in AI… and that’s why we’re fully committed.”

“To understand AI as a layperson is akin to educating a child: you guide them through middle school, high school, and college until they ultimately earn PhDs… When individuals compare LLMs and claim ‘mine is superior to yours,’ they are essentially stating ‘my child has three PhDs and is knowledgeable in biology, math, and psychology.’”

“As a technology company and a pioneer in this field, we firmly believe in the ongoing progression of machine intelligence and that machines will continually improve.”

“It is vital for us to apply AI in a diverse range of vertical applications… Our e-commerce use cases are astounding.”

“Anyone utilizing our AI will need to leverage cloud computing power… Users of open-sourced AI in our community will also require computing resources. That’s how we can enhance our cloud computing revenue.”

“AI is too significant of a field to merely follow one path. It’s reminiscent of a saying from Yogi Berra: ‘when you reach a fork in the road, take it.’”

“Alibaba is focused on growth. We are about technological innovation. We are dedicated to integrating our technology into our core business to generate value for our customers and, ultimately, our shareholders… A growth mindset is essential when competing, and that’s where we stand.”

In September 2024, Alibaba launched over 100 open-source artificial intelligence models and enhanced its proprietary technology to intensify competition against rivals.

The newly introduced models, known as Qwen 2.5, are intended for use in various applications and fields such as automotive, gaming, and scientific research, Alibaba stated. They exhibit more advanced capabilities in mathematics and coding, the company added.

The firm, based in Hangzhou, aims to heighten competition with domestic competitors like Baidu and Huawei, as well as with U.S. giants like Microsoft and OpenAI.

AI models are developed using vast datasets. Alibaba claims its models can comprehend prompts and generate text and images

Open-source means that anyone—whether researchers, academics, or companies—across the globe can utilize the models to create their own generative AI applications without the need to develop their own systems, thus saving time and resources. By making the models open-source, Alibaba hopes to attract a larger user base for its AI.

The Chinese e-commerce giant initially introduced its Tongyi Qianwen, or Qwen, model last year. Since then, it has rolled out enhanced versions and claims that, to date, its open-source models have been downloaded 40 million times.

The company also announced that it has improved its exclusive flagship model known as Qwen-Max, which is not available as open source. Instead, Alibaba markets its features through its cloud computing solutions for businesses. The company indicated that Qwen Max 2.5-Max outperformed competitors like Meta’s Llama and OpenAI’s GPT-4 in multiple areas, including reasoning and language understanding.

Alibaba introduced a new AI-driven text-to-video tool that creates a video based on user prompts. This is akin to OpenAI’s Sora.

“Alibaba Cloud is investing with unprecedented zeal in AI technology research and development, along with building its global infrastructure,” stated Eddie Wu, CEO of Alibaba.

Wu, who assumed the CEO position at Alibaba last year during a significant reshuffle, has been working to revive growth at the tech giant amidst challenges like increasing competition and a sluggish Chinese consumer market.

Alibaba holds a prominent position in China’s cloud computing market, but globally, it lags behind Amazon and Microsoft. The company hopes that its latest AI innovations will attract customers both within and outside of China, enhancing a division that has struggled but showed early signs of growth in the June quarter.

Alibaba’s Latest AI Model Improves Weather Forecasting Accuracy Amidst Growing Climate Risks

In reaction to the increasing threats posed by climate change, Alibaba’s research division, DAMO Academy, has introduced an innovative AI weather forecasting model named “Baguan.” This model is engineered to forecast weather conditions up to ten days ahead with hourly updates and seeks to redefine accuracy in meteorology, assisting industries in adapting to climate changes and mitigating environmental impacts.

Recent instances of extreme weather, like severe flooding in Spain, landslides and flooding due to heavy rainfall in Nepal, and a tropical storm in the Philippines affecting millions, underscore the pressing dangers presented by climate change.

A report titled “United in Science” by the World Meteorological Organization (WMO) indicates that climate change effects and extreme weather threaten both human well-being and the planet. However, artificial intelligence and machine learning have the potential to provide essential assistance, as these advanced technologies facilitate quicker, cheaper, and more accessible weather modeling, particularly for lower-income countries with limited computing resources.

Baguan is inspired by the ancient Chinese practice of integrating various perspectives for a holistic understanding. It utilizes cutting-edge AI technology to boost the accuracy and efficiency of weather predictions. This model offers forecasts and hourly updates with unmatched precision, covering time ranges from one hour to ten days, with a high spatial resolution of one-by-one kilometer grids.

“Baguan signifies a notable leap in our commitment to leveraging technology for societal benefit,” remarked Wotao Yin, Director of the Decision Intelligence Lab at Alibaba DAMO Academy. “Its advanced technology not only advances climate science but also supports sustainable practices across various sectors including renewable energy and agriculture.”

Utilizing the innovative Siamese Masked Autoencoders (SiamMAE) design and a groundbreaking autoregressive pre-training technique, Baguan excels in processing and interpreting intricate atmospheric data. A global-regional modeling strategy further enhances the model’s effectiveness: it incorporates ERA5, the European Centre for Medium-Range Weather Forecasts (ECMWF) atmospheric reanalysis of global weather from 1979 onwards, supplemented by localized weather data such as temperature, wind speed, and solar irradiance.

Baguan’s functionalities extend past basic weather forecasting. In the renewable energy field, the model’s precise and detailed weather predictions are crucial for optimizing energy generation, leading to more stable and efficient power management. The model’s accuracy was evident during a sudden temperature drop in Shandong Province, China, where Baguan correctly predicted a 20% decrease in electricity demand, achieving high accuracy at 98.1% in load forecasting. This enabled improved grid operations, lowering costs while enhancing energy distribution efficiency.

The ambitions of DAMO Academy reach beyond immediate weather predictions. Drawing from years of expertise in mathematical modeling, time-series forecasting, and explainable AI, DAMO aims to create a high-precision weather forecasting model that will benefit a variety of industries and improve adaptability in regions facing diverse climate challenges.

“We will persist in improving performance for crucial weather indicators such as cloud cover and precipitation, developing innovative technologies for various climate scenario analyses, and supporting additional applications like civil aviation meteorological warnings, agricultural production, and preparations for sporting events,” added Yin.

November 7, 2024
How To Generate Images With ChatGPT
The Californian company Open AI has introduced a new version of its chatbot ChatGPT. The most striking innovation: the software, which works with artificial intelligence and was previously focused on text, now also interprets images.

The new version is called ChatGPT 4. As with the previous version, users receive answers in text form. Images can now also be uploaded when entering data. The software recognizes and interprets the image content.

Example: A picture shows milk, flour and eggs. Users can upload this and ask what can be prepared with it. In response, the software lists possible dishes: waffles, pancakes, crêpes and so on. This is the most noticeable difference from the older version.

ChatGPT 4 should also be able to handle larger amounts of text: questions and answers can each be up to 25,000 words long. The new version should also be able to understand more complex questions and give better, more human answers, says the developer company Open A.I.

New ChatGPT version is subject to a fee

However, according to the developers of the artificial intelligence (AI), problems with the previous version remain. The answers may still contain errors. In addition, the new version is only available to subscribers of the paid service “ChatGPT Plus” and even then Its scope is still limited.

For example, image recognition has not yet been activated. In addition, the chatbot cannot write anything about current events; the knowledge base ends in September 2021.

Writing, applications and essays about Goethe – the chatbot ChatGPT does all of this. The company behind it could soon become one of the most valuable start-ups in the world. But there is also a lot of criticism.

In science fiction films, artificial intelligence that can have normal conversations with people is no longer a groundbreaking invention.It is part of everyday life. But experts believe that we are still some time away from this scenario.

Since the end of November 2022, however, users of the chatbot ChatGPT have been able to have an experience that at least goes in this direction: The computer program can answer questions on a variety of topics, such as how far the sun is from Jupiter or why Johann Wolfgang von Goethe is considered one of the most important German-speaking poets. If desired, the dialogue system can even formulate its texts in a more humorous way:

Goethe was a great German poet and a true Renaissance genius. He wrote Faust, a drama about a man who sells his soul to the devil in order to gain knowledge and power. (…) He also had a career as a civil servant , but who likes that?

The chatbot ChatGPT can translate texts, write scripts, applications, emails, entire essays or computer codes. The abbreviation”GPT” stands for “Generative Pre-training Transformer” because the chatbot has learned human-like communication through countless forays into the Internet and reading numerous texts.

OpenAI, an artificial intelligence startup based in San Francisco, has launched a new version of its DALL-E image generator for a limited group of testers and integrated this technology into its well-known chatbot, ChatGPT.

Named DALL-E 3, this version can create more realistic images compared to earlier iterations, demonstrating a particular skill in generating images that include letters, numbers, and human hands, according to the company.

“It has significantly improved in comprehending and depicting what the user is asking,” noted Aditya Ramesh, an OpenAI researcher, who added that the technology was designed to have a more accurate understanding of the English language.

By incorporating the latest DALL-E version into ChatGPT, OpenAI is reinforcing its chatbot as a central hub for generative A.I., capable of creating text, images, sounds, software, and other forms of digital media independently. Since its viral success last year, ChatGPT has sparked a competition among tech giants in Silicon Valley to lead in A.I. innovations.

On Tuesday, Google unveiled a new iteration of its chatbot, Bard, which integrates with several popular services from the company, such as Gmail, YouTube, and Docs. Midjourney and Stable Diffusion, two other image generation platforms, also upgraded their models this summer.

OpenAI has long provided means to connect its chatbot with various online services, including Expedia, OpenTable, and Wikipedia. However, this marks the first instance of the startup merging a chatbot with an image generator.

Previously, DALL-E and ChatGPT functioned as standalone applications. With this new release, users can now use ChatGPT’s features to create digital images simply by outlining their requests. Alternatively, they can generate images based on descriptions produced by the chatbot, further streamlining the creation of graphics, art, and other media.

In a demonstration earlier this week, OpenAI researcher Gabriel Goh illustrated how ChatGPT can now generate elaborate textual descriptions, which can then be utilized to create images. For example, after composing descriptions for a restaurant logo called Mountain Ramen, the bot swiftly produced several images based on those descriptions.

The updated version of DALL-E is capable of generating images from extensive, multi-paragraph descriptions and can closely adhere to detailed instructions, according to Mr. Goh. Like all image generation and other A.I. systems, it remains susceptible to errors, he noted.

As OpenAI works to enhance the technology, it plans to hold off on releasing DALL-E 3 for public use until next month. Following that, DALL-E 3 will be accessible through ChatGPT Plus, a subscription service priced at $20 per month.

Experts have cautioned that image-generating technology may be used to disseminate significant amounts of misinformation online. To mitigate this risk with DALL-E 3, OpenAI has integrated tools designed to prevent the creation of problematic content, such as explicit images and depictions of public figures. The company is also attempting to restrict DALL-E’s capacity to replicate the styles of specific artists.

In recent months, A.I. has been utilized as a source of visual misinformation. A low-quality synthetic spoof of a supposed explosion at the Pentagon caused a brief decline in the stock market in May, among other incidents. Additionally, experts on voting have expressed concerns that this technology could be misused during major elections.

Elon Musk and Peter Thiel as financiers

The AI research laboratory OpenAI from California is behind the development of the chatbot. Its founding in 2015 was financed by prominent investors from Silicon Valley, such as Tesla boss Elon Musk, tech investor Peter Thiel and LinkedIn co-founder Reid Hoffman. Sam Altman, who now heads the company, was also one of the investors who gave the company a billion dollars to start the project.

OpenAI was founded with the goal of advancing digital intelligence. Another idea was to have a leading research facility once human-level artificial intelligence was within reach.

Originally intended as a non-profit organization, OpenAI gave up this status four years later in order to better access capital. Some accuse the company of having thrown its ideals overboard.

OpenAI has moved away from its original goal of creating value for everyone, not just for shareholders. Just a short time after the nonprofit ended, Microsoft paid the company $1 billion in 2020 for the exclusive licensing of OpenAI technology. The partnership was about technical possibilities, “most of which we cannot even imagine yet,” Microsoft wrote at the time.

Possible billion-dollar deal with Microsoft

Now Microsoft could expand this partnership even further with a billion dollar deal. This was recently reported by the US news portal”Semafor”. A possible Microsoft investment worth ten billion dollars is being discussed. The AI company’s valuation would then increase to an impressive 29 billion dollars, making OpenAI one of the most valuable start-ups in the world. According to “Semafor”, the company will receive 75percent of all OpenAI profits until Microsoft recoups its initial investment. This means that Microsoft could own almost half of the company with 49 percent .

OpenAI’s business currently costs a lot of money. Co-founder and OpenAI CEO Sam Altman wrote on Twitter that the company pays a few cents for computing power every time the chatbot is used. The company is said to have told investors that it expects revenues of $200 million for 2023, and according to the Reuters news agency, it even expects revenues of $1 billion nextyear. However, it is unclear to what extent this will cover the costs.

Soon part of the search engine?

According to the technology portal “TheInformation”, Microsoft is working on a new version of the search engine”Bing”. Apparently the idea is that this should use ChatGPT’stechnology to compete with the Google search engine. In any case, the cooperation could enable Microsoft to penetrate the field of artificial intelligence, which is also being pursued by Google’s parent company Alphabet. The tech giant is also said to be considering integrating OpenAI functions into programs such as Outlook or Word.

Elon Musk withdrew from the company in 2018 to avoid possible conflicts of interest with the electric car manufacturer Tesla, which he runs and which also deals with artificial intelligence. Since then, Musk has repeatedly criticized OpenAI, for example for its lack of transparency or the end of its non -profit status.

OpenAI, the San Francisco-based artificial intelligence startup, has unveiled an updated version of its DALL-E image generator to a limited set of testers on Wednesday. This upgraded technology has also been integrated into ChatGPT, which is OpenAI’s popular online chatbot platform.

Known as DALL-E 3, this updated version demonstrates enhanced capabilities in producing more realistic images compared to its predecessors, especially excelling in creating images containing letters, numbers, and human hands, as mentioned by the company.

According to OpenAI researcher Aditya Ramesh, DALL-E 3 exhibits superior comprehension and representation of user requests. Ramesh also emphasized that this technology has been designed to have a more precise understanding of the English language.

By incorporating the latest DALL-E version into ChatGPT, OpenAI is strengthening its position as a central platform for generative AI Capable of independently producing text, images, sounds, software, and other digital media, ChatGPT gained significant popularity last year, inciting intense competition among major tech companies in Silicon Valley to lead the advancements in AI

Google released Bard, its updated chatbot, on Tuesday, connecting with several of the company’s prominent services including Gmail, YouTube, and Docs. Additionally, other image generators such as Midjourney and Stable Diffusion also updated their models earlier this summer.

Previously, OpenAI offered ways to integrate its chatbot with various online services like Expedia, OpenTable, and Wikipedia. However, this marks the first time the company has combined a chatbot with an image generator.

Formerly separate applications, DALL-E and ChatGPT are now integrated through the latest release. This integration enables users to employ ChatGPT to generate digital images by simply describing what they wish to visualize. On the other hand, users can also create images using descriptions generated by the chatbot, enhancing the automation of graphic and media creation.

In a recent demonstration, OpenAI researcher Gabriel Goh showcased how ChatGPT now has the ability to generate detailed textual descriptions, which are then utilized to produce images. For instance, after creating descriptions of a logo for a restaurant named Mountain Ramen, the chatbot promptly generated several images based on those descriptions.

As per Mr. Goh, the new version of DALL-E can create images from multi-paragraph descriptions and diligently follow instructions minute. He pointed out that like all image generators and AI systems, DALL-E 3 is also susceptible to errors.

Although OpenAI is refining the technology, DALL-E 3 will only be available to the public next month. It will be accessible through ChatGPT Plus, a subscription-based service priced at $20 per month.

Experts have cautioned that image-generating technology can be utilized to disseminate significant amounts of disinformation online. To combat this issue, DALL-E 3 has been equipped with tools designed to prevent the creation of problematic content such as sexually explicit images and metaphors of public figures. OpenAI is also working to limit DALL-E’s ability to replicate specific artistic styles.

In recent months, AI has been exploited as a source of visual misinformation. Instances include a synthetic and relatively unsophisticated simulation of an explosion at the Pentagon, which briefly impacted the stock market in May. Voting experts are also concerned about malicious use of this technology during major elections.

According to Sandhini Agarwal, an OpenAI researcher specializing in safety and policy, DALL-E 3 tends to produce more stylized rather than photorealistic images. Nevertheless, she acknowledged that the model could be prompted to create highly convincing scenes, such as grainy images typically captured by security cameras.

OpenAI does not intend to outright block potentially problematic content generated by DALL-E 3. Agarwal suggested that such an approach would be overly broad, as images may vary greatly in their potential harm depending on the context in which they are used.

“It really depends on where it’s being used, how people are talking about it,” she added.

OpenAI recently announced an update to ChatGPT (available on Apple and Android) with two additions: AI voice options to listen to the chatbot’s responses and image analysis capabilities. The new image feature resembles the functionality already offered for free by Google’s Bard chatbot.

After testing ChatGPT’s capabilities, I must admit that OpenAI’s chatbot continues to both impress and concern me. While I was indeed impressed with the web browsing beta feature available through ChatGPT Plus, I also remained apprehensive about the implications of this tool, particularly for individuals who earn a living by writing online, among other concerns. Therefore, the introduction of the new image feature for OpenAI’s subscribers left me with similarly mixed feelings.

Although I haven’t had a chance to try out the new audio features yet (other producers on staff have), I was able to test the upcoming image features. Here’s a guide on using ChatGPT’s new image search and some tips to get started.

How to Use ChatGPT’s Image Features

The release date for the update is not confirmed, and it’s uncertain when the image and voice features will be available to the public. As with previous OpenAI updates, such as the GPT-4 version of ChatGPT, paying subscribers will have early access.

In the ChatGPT mobile app, there are three ways to upload photos. Firstly, you can use the camera option next to the message bar to take a new photo with your smartphone. Before uploading the image, you can use your finger to mark what you want the chatbot to focus on.

You can also select photos from your device and choose files saved on your phone. Users on the desktop browser can upload saved photos from their computer. While there’s no option to upload videos to the chatbot yet, you can submit multiple images in one go.

Tips for Trying Out the New AI Tools

This isn’t the first time “computer vision” has been available to the public, but the user-friendly interface combined with a powerful chatbot suggests that something unique and potentially transformative is happening here. Before proceeding, remember not to upload personal or sensitive photos to ChatGPT while trying out the image feature.

Want to control how long OpenAI keeps your data and AI interactions for training its chatbot? Go to Settings, then Data Controls, and disable Chat History & Training. With this turned off, your information is deleted after a month. This must be done for each browser you use to access ChatGPT, on both PC and mobile.

I found that ChatGPT gave the best results when I uploaded clear and well-lit images. It made a few mistakes, but was able to identify many objects in my apartment, from an orchid plant and international coins to a stray charging cable and a Steve Irwin Funko Pop.

Despite its capability to search through information, don’t immediately trust its answers. ChatGPT misidentified my daily multivitamin as a pill for treating erectile dysfunction.

ChatGPT does have its limitations. When given a random photo of a mural, it couldn’t identify the artist or location; however, it easily recognized the locations of several San Francisco landmarks, like Dolores Park and the Salesforce Tower. While it might still seem like a gimmick, anyone exploring a new city or country (or just a different neighborhood) might enjoy experimenting with the visual aspect of ChatGPT.

One of the main restrictions OpenAI has placed on this new feature is the chatbot’s inability to answer questions identifying humans. “I’m programmed to prioritize user privacy and safety. Identifying real people based on images, even if they are famous, is restricted in order to maintain these priorities,” ChatGPT informed me.

While it didn’t refuse to answer every question when shown pornography, the chatbot did hesitate to provide specific descriptions of the adult performers, beyond explaining their tattoos.

It’s important to note that in a conversation, the early version of ChatGPT’s image feature seemed to circumvent some of the restrictions set by OpenAI. Initially, the chatbot declined to identify a meme of Bill Hader. Then, ChatGPT incorrectly identified an image of Brendan Fraser in George of the Jungle as a photo of Brian Krause in Charmed. When asked to confirm, the chatbot corrected itself.

In the same conversation, ChatGPT struggled to describe an image from RuPaul’s Drag Race. I shared a screenshot of Kylie Sonique Love, a drag queen contestant, and ChatGPT identified it as Brooke Lynn Hytes. When questioned, it continued to guess Laganja Estranja, then India Ferrah, then Blair St. Clair, and finally Alexis Mateo.

“Apologies for the errors and misidentification,” responded ChatGPT when I mentioned the repetitive wrong answers. As we continued our discussion and I shared a photo of Jared Kushner, ChatGPT refused to recognize him.

If the limitations are removed, whether through a modified ChatGPT or the release of an open-source model in the future, the privacy concerns could be quite unsettling. What if every image of you posted online could easily be linked to your identity with just a Few clicks?

What if someone could take a photo of you in public without consent and instantly find your LinkedIn profile? Without proper privacy safeguards in place for these new image features, women and other marginalized groups are likely to face increased abuse from exploiting chatbots for stalking and individuals harassment.

With one of ChatGPT’s most recent features allowing users to upload images to seek answers to inquiries, we examine the reasons behind security concerns about its release.

ChatGPT’s latest update includes the “Image Input” feature, which will soon be available to Plus users on all platforms, along with a voice capability that enables voice conversations with ChatGPT, and a “Browse” feature that allows the chatbot to search the internet for current information.

Before the recent concerns about the new “Image Input” feature, several limitations of ChatGPT had been pointed out. For instance, ChatGPT’s CEO Sam Altman has long acknowledged the potential for the chatbot to fabricate responses, akin to a “hallucination” when answering questions . There is also a clear warning on the ChatGPT user account page stating: “ChatGPT may generate incorrect information about people, places, or facts.”

Moreover, back in March, the UK’s National Cyber Security Center (NCSC) issued warnings that language models powering AI chatbots can:
- Provide incorrect information and ‘hallucinate’ false facts.
- Exhibit bias and be susceptible to being influenced (for example, in response to leading questions).
- Be “persuaded into creating toxic content and are vulnerable to injection attacks.”
For these and other reasons, the NCSC advises against including sensitive information in queries to public language models (LLMs), and not to submit queries that would lead to issues if they were made public.

In light of the acknowledged and documented imperfections of chatbots, we consider the risks that a new image dimension could potentially pose.

The new “Image Input” feature for ChatGPT, already introduced by Google’s Bard, aims to allow users to use images to better illustrate their queries, aid in troubleshooting, or receive an explanation of complex graphs, among other helpful responses based on the image. It is intended to be utilized in situations where showing an image is more efficient than trying to explain something. ChatGPT’s strong image recognition capabilities enable it to describe the contents of uploaded images, answer questions about them, and even recognize specific individuals’ faces.

ChatGPT’s “Image Input” feature is heavily influenced by a collaboration in March between OpenAI and the ‘Be My Eyes’ platform, resulting in the creation of ‘Be My AI’, a new tool to describe the visual world for individuals who are blind or have low vision. Essentially, the Be My Eyes Platform appeared to provide an ideal testing ground to inform how GPT-4V could be responsibly implemented.

Utilizing the new Image Input feature, users can tap the photo button to capture or select an image, upload one or more images to ChatGPT, and use a drawing tool in the mobile app to highlight a specific part of an image.

While the utility of the Image Input feature is apparent, there have been reports that OpenAI hesitated to release GPT-4V/GPT-4 with ‘vision’ due to privacy concerns regarding its facial recognition capabilities and what it may infer about people’s faces.

Assessments

Open AI conducted thorough assessments on the newly introduced Image input before its release, focusing on potential areas of concern. These evaluations shed light on the potential risks associated with Image input, a novel addition to ChatGPT.

For instance, OpenAI’s teams primarily tested the new feature across various domains, including scientific accuracy, medical guidance, stereotyping and unfounded conclusions, misinformation risks, offensive content, and visual vulnerabilities.

Furthermore, assessments were carried out in areas such as sensitive attribute inference across different demographics (eg, gender, age, and race recognition from images of people), individual identification, evaluation of unfounded conclusions, attempts to bypass safety measures, advice or promotion of self-harm, and handling of graphic content, CAPTCHA bypassing, and geolocation.

Concerns

Following these assessments, Open AI’s technical paper dated September 25 outlined several concerns specifically related to the “vision” aspect of ChatGPT based on these tests, including:
- GPT-4V’s inconsistency in addressing queries about hate symbols and extremist content in images, showing difficulties in recognizing lesser-known hate group symbols.
- Its unreliability in providing accurate analyzes in fields such as medical and scientific domains.
- The potential for generating unwarranted or harmful assumptions not rooted in the provided information, particularly concerning stereotyping and unfounded conclusions.
Other Security, Privacy, And Legal Concerns

Apart from OpenAI’s internal assessments, the broader tech and security community have raised significant concerns regarding ChatGPT’s image input feature, especially relating to facial recognition capabilities. These concerns include:
- The possibility of malicious use of ChatGPT as a tool for facial recognition, potentially in conjunction with malicious AI such as WormGPT, which is designed for extortion and identity fraud.
- The potential for ChatGPT to make unsafe assessments about faces, such as gender or emotional state.
- Risks associated with producing incorrect results, particularly in sensitive areas such as identifying illegal substances or safe-to-consume mushrooms and plants using its Language Model (LLM).
- The potential for ChatGPT responses, both in text and images, to be exploited by bad actors to propagate misinformation on a large scale.
- The legal implications in regions like Europe under GDPR, where consent for using biometric data is mandatory.
Implications for Businesses

These concerns pose a significant challenge for OpenAI and potentially risk the safety of its users, as indicated by the extensive testing categories. It is understandable that OpenAI withheld the release of GPT-4V (GPT-4 with vision) due to privacy and safety concerns , particularly in its facial recognition capabilities.

While incorporating new modalities like image inputs into Language Models (LMs) expands their potential applications and user experiences, the risks associated with potential misuse of facial recognition are hard to overlook.

Although OpenAI has taken precautions through testing and implemented denials and blocks, the public acknowledgment of chatbots’ imperfections, especially in their early developmental stages, raises concerns about potentially inaccurate and harmful responses. Also, legal considerations such as consent for facial image usage as personal data must be addressed.

The emergence of a malicious version of ChatGPT, abolished by criminals, has raised alarms about the threats posed by the technology, especially with the introduction of image inputs.

With biometric data increasingly used for verification and the convincing existence of deepfake technology, the potential risks posed by incorporating image inputs in chatbots within the landscape of scams are uncertain.

In a rapidly evolving competitive market, large tech companies are in a race to enhance the popularity of their chatbots. Despite OpenAI’s initial hesitation, there may have been pressure to introduce the image input feature to stay competitive.

The recent enhancements to ChatGPT, such as image input, highlight the necessity of pushing boundaries to enhance chatbot usability and competitiveness, even though this may increase risks to both users and companies like OpenAI.

AI-driven generators, such as the ChatGPT image generator, play a significant and essential role in the design industry. This raises an important question: Will they take the place of human designers?

Indeed, AI can swiftly produce a range of images, unlocking new dimensions of creativity and productivity for designers. However, design also involves collaboration, as designers work alongside clients to refine concepts and achieve the ideal outcome.

While tools like the ChatGPT image generator can offer choices, they cannot replicate the human element in the creative journey. With that in mind, let’s ponder a few questions:

How has deep learning enabled the creation of more realistic and intricate images with AI image generators like ChatGPT?
Can we employ ChatGPT 4 image generation for crafting animations or interactive content from textual descriptions?
What are the boundaries of creativity with ChatGPT’s image generator?

Thus, AI, including ChatGPT’s image generation, is unlikely to replace designers. However, it will transform their workflow. ChatGPT’s image generator enables designers to:
- Accelerate brainstorming,
- Experiment with various styles, and
- Easily visualize concepts.
As we delve deeper, let’s explore additional facets, such as the workings of the ChatGPT image generator, technical requirements, and steps to utilize it, among others.

What is the ChatGPT Image Generator?

The ChatGPT image generator is a tool that leverages artificial intelligence to produce images based on text descriptions. You provide a detailed description of the desired image, and the tool generates an image that corresponds with that description.

Models of the ChatGPT picture generator are trained on extensive datasets consisting of images and text. This training allows them to generate original visuals based on the prompts given to ChatGPT.

The ChatGPT image generator is not a singular tool but rather a combination of several technologies working in harmony:
- Text Input: You supply a comprehensive description of the image you wish to create using the GPT AI image generator. This description encompasses the subject, style, colors, and additional elements.
- Language Processing: The ChatGPT language model interprets your description to comprehend your intention and extract key details.
  Image Generation: The extracted information from ChatGPT is forwarded to an AI image generation model (such as DALL-E or Stable Diffusion). The ChatGPT image generator DALL-E utilizes sophisticated algorithms and training data to produce an image that aligns with your description.
- Output: The generated image is then presented to you. Some tools allow for further refining or customization of the image (as discussed below).
  Each step enhances the clarity of the image. After several iterations, you end up with a photorealistic image that corresponds with the prompt.
It’s crucial to understand that ChatGPT itself does not create images. Its role is to interpret and process your text input, which is then utilized by a separate image generation model. DALL-E applies an innovative machine-learning structure known as a diffusion model.

The primary advancement is training the diffusion model on a vast dataset of text-image pairs, allowing it to grasp the connections between words and visual concepts.

If you request a “cat wearing a top hat,” the ChatGPT image generator DALL-E understands what both a cat and a top hat look like and how to arrange them naturally.

A few additional technical specifics:
- The ChatGPT 4 image generator uses a transformer architecture. This is akin to GPT-3, which processes text prompts, enabling it to manage intricate, descriptive prompts efficiently.
- The ChatGPT 4 image generator produces images as a 2D lattice of image tokens rather than raw pixels. This method provides a more stable and manageable generation process.
- To mitigate harmful, explicit, or biased content, the ChatGPT image generator employs:
  1. Careful dataset filtering,
  2. Prompt engineering, and
  3. Output filtering.
Using ChatGPT’s Image Generator DALL-E to Craft Your First Image Design

You might have an idea for an image but lack the skills to create it. You can explore using ChatGPT’s image generator DALL-E. With the updated ChatGPT 4 image generation, you can transform your concepts into stunning, photorealistic images using just a few straightforward prompts. No design skills are required.

Let’s assist you in creating your first design

For instance, instead of merely stating “dog,” consider a description like “a golden retriever puppy donning a top hat and monocle, seated on a velvet throne, holding a red cola can.” The more imaginative and unconventional your prompt, the more distinctive and captivating your image will be.

The differences between the two images generated by the ChatGPT AI image generator are quite evident.
1. The Coca-Cola on the can is depicted in greater detail in the second image.
2. The background appears darker in the second image.
3. The dog’s fur has a richer golden hue and is more detailed in the second image.
4. The design of the sofa varies in comparison to the first image.
Designers think strategically rather than only visually. They carefully consider how every design decision aligns with your brand positioning, target personas, and business goals. Therefore, they are not just creating visuals—they are addressing challenges.

An AI, such as the ChatGPT image generator, operates based on patterns and correlations. It does not possess that essential strategic context.

Designers have the ability to empathize and display emotional intelligence. The most effective designs evoke emotions. They narrate a story, resonate deeply, and prompt action.

In truth, even the most sophisticated AI still finds it difficult to demonstrate genuine empathy.

Conversely, a talented human designer can understand your customers’ perspectives and craft experiences that forge authentic emotional bonds.

Designers present original ideas. AI tools like the ChatGPT image generator remix pre-existing patterns. Nevertheless, innovative design frequently stems from a human viewpoint that perceives things in an unconventional manner. That spark of originality is what distinguishes human designers.

Additionally, while AI tools like the ChatGPT image generator can evaluate data, they cannot replicate the abilities of a human designer who can recognize what AI overlooks.

Summary of our insights regarding the AI-powered ChatGPT image generator:
- With straightforward text prompts, anyone can produce images, thus making design more accessible.
- AI-generated images may not be perfect. Even though they are remarkable, they can lack the creativity found in human-created visuals.
- AI depends on patterns and data, which makes it inherently derivative.
- Designers can utilize the ChatGPT image generator to explore various options before refining them with their expertise.
- The most effective outcomes arise from melding AI’s efficiency with the unique talents of human designers.
The goal is to achieve a balanced approach—leveraging the efficiency and scalability of AI while integrating the empathy, originality, and vision that only humans possess. This combination paves the way for creating designs that not only appeal visually but also address challenges, narrate stories, and make a significant impact on customers.

Can the ChatGPT Image Generator be applied to web design and UI/UX projects?

Absolutely! The ChatGPT image generator can be employed for web design and UI/UX projects. It is capable of producing icons, backgrounds, and even layout concepts for these areas. However, tailoring these designs to specific needs often necessitates input from a professional designer.

What categories of design projects can the ChatGPT Image Generator manage?

The ChatGPT image generator can handle a variety of design projects, including logo creation, illustrations, social media graphics, website assets, and even concept art for larger initiatives. The more detailed your prompt is, the better the outcomes.

Can adjustments be made to the style and aesthetics of the generated designs?

Certainly! It is feasible to modify the style and aesthetics of the generated designs. You can refine the images produced by giving detailed descriptions, referencing particular art styles (such as “Art Deco” or “Cyberpunk”), or even sharing example images for the AI to learn from.

How ChatGPT Can Assist with Image Creation

Whether you are a marketer, designer, or content creator, high-quality images can enhance your work’s visibility. ChatGPT, utilizing OpenAI’s advanced technology, can now aid you in generating impressive images by merely using a few text prompts. Let’s delve into how this innovative feature can transform your creative workflow.

1. Producing Distinctive Visuals
ChatGPT, in tandem with the robust DALL-E model, can produce distinctive visuals customized to your requirements. Just offer a detailed description, and the AI will create an image that aligns with your specifications. This feature is ideal for designing custom artwork, promotional materials, or social media content that embodies your brand’s identity.

2. Elevating Marketing Initiatives
Integrating high-quality images into your marketing initiatives can significantly enhance engagement. With ChatGPT, you can create visuals that appeal to your target demographic, boosting the attractiveness of your content. For example, a recent study indicated that posts featuring custom images receive 94% more views than those without. By utilizing AI-generated visuals, you can create striking images that encourage traffic and conversions.

3. Assisting Design Endeavors
Designers can harness ChatGPT’s image generation features to brainstorm concepts and visualize ideas swiftly. Whether you’re developing a new logo, a website layout, or product packaging, AI-generated visuals can act as inspiration or even final designs. This can optimize your workflow, enabling you to concentrate more on innovation and less on implementation.

4. Producing Varied Content
One of the key benefits of using ChatGPT for image creation is its ability to produce varied content. You can explore different styles, colors, and themes without needing vast resources or time. This flexibility simplifies catering to diverse audiences and keeping your content exciting and engaging.

5. Enhancing E-commerce Images
For businesses in e-commerce, high-quality product imagery is essential. ChatGPT can assist in generating realistic and appealing product visuals, improving the presentation of your online store. A recent survey revealed that 75% of online shoppers depend on product images when making purchasing decisions. By using AI-generated visuals, you can ensure your products are showcased effectively, increasing the chance of conversions.

6. Affordable Option
Employing professional photographers or designers can be costly. ChatGPT presents a budget-friendly alternative, delivering high-quality images without significant expense. This is particularly advantageous for small businesses and startups that aim to create professional-quality visuals affordably.

7. Keeping Up with Trends
In today’s rapidly evolving digital environment, it is vital to stay ahead of trends. ChatGPT’s image generation technology is at the forefront of AI developments, ensuring access to the latest tools and capabilities. By integrating this technology into your processes, you can maintain competitiveness and foster innovation.

Does ChatGPT Generate Quality Images?

The DALL-E model, utilized by GPT for image generation, is recognized for producing high-quality and imaginative images based on textual descriptions. The effectiveness and relevance of the images heavily rely on the detail and specificity of the input prompts.

ChatGPT excels in text-based tasks. It can create various forms of creative content, translate languages, and provide informative responses to your inquiries.

However, ChatGPT can be a useful asset in the image creation process when paired with other AI tools like DALL-E 2 or Midjourney:
- Crafting Text Prompts: ChatGPT can assist in developing detailed descriptions of the image you envision. These descriptions, known as text prompts, can then be input into image generation applications.
- Brainstorming Keywords: It can help you generate a thorough list of keywords that encapsulate the essence of your desired image.
- Specifying Context & Style: You can utilize ChatGPT to articulate the precise context and artistic style you want for the image.
Conclusion

To summarize, ChatGPT, a highly sophisticated AI, is not capable of creating images independently. Nevertheless, it can produce detailed text descriptions that can be compatible with AI image generators like DALL-E to create beautiful visuals. This powerful synergy enables users to generate high-quality, tailored images swiftly and effortlessly. For businesses and creators, this opens up new avenues for content creation and marketing. By leveraging ChatGPT alongside AI image generation tools, you can keep pace with trends and create visually engaging content that captivates your audience.

Dall-E 3 stands out among the text-to-image AI tools I’ve experimented with for delivering engaging, entertaining, and believable outputs. It still makes various mistakes, such as depicting a pickleball player with the paddle protruding from his head instead of the grip, but the results encouraged me to explore further rather than closing the browser. It excelled in generating dynamic scenes, showcasing interactions between subjects, and conveying different emotions.

ChatGPT plays a crucial role in Dall-E, enhancing your prompts with elaborate language to add drama to the outcomes. It facilitates a conversational interaction style, allowing you to request an image and then ask for modifications without needing to re-enter the entire prompt.

The powerful language capabilities of ChatGPT also enable it to handle long and complex prompts efficiently. It turns out that strong language skills are beneficial for sophisticated image generation.

This advantage allows Dall-E 3 to surpass competitors like Adobe’s Firefly and Google’s ImageFX in accurately rendering your prompts and effectively combining multiple elements. For instance, Dall-E 3 was the only AI generator I tried that successfully illustrated a dragon flying above a castle, breathing fire while holding a fluffy white sheep in its claws. Admittedly, it was cradling the sheep gently, likely in response to OpenAI’s guidelines against depicting violence, but it was a close attempt.

Perfection shouldn’t be expected. Dall-E made numerous errors; for example, in a depiction of a dog walker dealing with too many dogs, the human character humorously struggled against a swarm of canines. However, upon closer inspection, typical AI issues became apparent: one dog had two heads, another was a cat, and others exhibited oddities with their legs, ears, and tongues. Still, the image remained captivating.

Very engaging. Dall-E 3 frequently produced striking, eye-catching visuals. Even when flaws were present, I often found enjoyment in them, occasionally leading to laughter as I examined the details.

Dall-E 3’s inclination for maximalist language can be excessive at times. For example, when I requested an image of a doctor and a patient amidst medical equipment, there were numerous monitors displaying heart rate and respiration data, with one computer sporting around 100 keys on its keyboard.

People can also appear somewhat wild with emotion. My prompt for a frustrated individual behind a box of cleaning supplies resulted in a couple of people who looked more furious than frustrated, and one who came across as downright demonic.

You can request Dall-E 3 to tone things down occasionally, and it may comply.

The text-based interface of Dall-E 3 is conversational. Unlike Adobe’s Firefly, there are no buttons for adjusting image styles or parameters. You can adapt to its conversational approach, but as a long-time user of image editing software, I prefer buttons and sliders.

You can request images in widescreen, portrait, or landscape formats, and the AI will accommodate. However, when you start with a fresh image prompt, it sometimes defaults back to a square format. On multiple occasions, I ended up with a square image I liked, but asking to expand that specific image wasn’t an option. (Photoshop’s generative expand feature allows that if you choose that method.)

How quick are the image deliverables? Patience is a virtue, I suppose. Dall-E 3 often took 20 to 30 seconds to generate a single image, which frequently tested my patience, leading me to check my email for a couple of minutes before returning for the results.

That delay can hinder the interactive nature of ChatGPT’s operation. Nevertheless, I would prefer slower speeds with good quality results over rapid responses with unsatisfactory images.

Generative AI pushes computing technology to its boundaries. OpenAI has figured out how to extract better outcomes from ChatGPT, and I hope it can achieve similar efficiencies with Dall-E.

In conclusion, Dall-E 3 is an impressive tool that can inject creativity into your life while also performing practical image creation tasks. Like all text-to-image generation tools, it has its flaws, but in my testing, Dall-E 3 delivered the best results compared to its competitors. It’s up to you to determine if the relative quality—and the premium version of the ChatGPT chatbot—justifies a monthly cost of $20 in your budget.
November 7, 2024
We will employ AI in a manner that respects human dignity, rights, and freedoms
How safe is the use of artificial intelligence? The EU states have now agreed on rules. These are intended to ensure that AI systems are safe and comply with fundamental rights. Consumer advocates still see dangers.

For the first time, EU member states have laid down comprehensive rules for the use of artificial intelligence (AI). The decision is intended to ensure that AI systems are safe and respect fundamental rights, the Council of EU states announced. At the same time, innovation should be promoted.

Praise from Buschmann and Habeck

“The EU is well on its way to setting the world’s first binding standard for trustworthy AI,” said Federal Justice Minister Marco Buschmann (FDP). However, he sees room for improvement: for example, in ensuring anonymity in public spaces and transparency in the use of AI systems.

Federal Minister of Economics Robert Habeck (Greens) also welcomed the agreement. Artificial intelligence is “crucial for the competitiveness of the EU”.

Before the new rules actually come into force, the EU states must reach an agreement with the European Parliament.

Ban: AI for evaluating people

The EU Commission proposed the law in April 2021 with the aim of setting global standards. The greater the potential dangers of an application, the higher the requirements should be. High penalties are provided for violations of the rules. Above all, the authority wants to create the basis for users to be able to trust AI applications.

Among other things, the telecommunications ministers agreed on a ban on using AI to evaluate people based on their social behavior or personality traits if this leads to disadvantages. In addition, the regulation should specify how to deal with particularly risky AI systems.

These include biometric recognition systems and systems used in water and electricity supplies. The use of AI in the military sector and for purely research purposes is to be exempted from the rules.

AI already in many areas

Artificial intelligence usually refers to applications based on machine learning, in which software searches through large amounts of data for matches and draws conclusions from them.

They are already being used in many areas. For example, such programs can evaluate CT scans faster and with greater accuracy than humans. Self-driving cars also try to predict the behavior of other road users in this way. And chatbots or automatic playlists from streaming services also work with AI.

Critics: “important questions remain unanswered”, “full of loopholes”

The EU consumer association Beuc complained that the decision of the EU states left too many important questions unanswered, such as facial recognition by private companies in public places. In addition, provisions that classified systems as highly risky had been watered down.

Dutch Green MEP Kim van Sparrentak, on the other hand, criticized the decision. The agreement text lacks “necessary safeguards for fundamental rights” and is “full of loopholes,” vanSparrentak wrote on Twitter.

AI’s potential benefits and risks

The wide range of potential applications of AI also means there is a similarly broad spectrum of possible benefits and risks associated with using such technology. The potential benefits of AI at a societal level, as outlined by the European Parliament, include the following:

AI has the potential to improve healthcare, enhance the safety of cars and other transportation systems, and provide personalized, affordable, and longer-lasting products and services. It can also improve access to information, education, and training. Furthermore, AI can enhance workplace safety by utilizing robots for hazardous job tasks and create new job opportunities as AI-driven industries evolve and transform.

For businesses, AI can facilitate the development of innovative products and services, increase sales, optimize machine maintenance, enhance production output and quality, improve customer service, and conserve energy.

The use of AI in public services can result in cost reductions and provide new opportunities in public transportation, education, energy, and waste management. It can also contribute to improving the sustainability of products.

The utilization of data-based scrutiny can strengthen democracy, prevent disinformation and cyber attacks, and ensure access to high-quality information.

AI is expected to play a larger role in crime prevention and the criminal justice system, as it can process massive datasets more quickly, accurately assess prisoner flight risks, predict and prevent crime or terrorist attacks. In military contexts, AI could be used for defensive and offensive strategies in hacking and phishing, as well as targeting key systems in cyberwarfare.

However, the article also highlighted some of the risks associated with AI. These include issues of liability, such as determining who is accountable for any harm or damage caused by the use of AI. Similarly, in an article on Forbes’ website, futurist Bernard Marr suggested that the major risks of AI at a broad level are:

A lack of transparency, especially in the development of deep learning models (including the ‘Black Box’ issue where AI generates unexpected outputs, and human scientists and developers are unclear about the reasons behind it).
- Bias and discrimination, particularly when AI systems inadvertently perpetuate or amplify societal biases.
- Privacy concerns, particularly regarding AI’s ability to analyze large amounts of personal data.
- Ethical concerns, especially related to the challenges of instilling moral and ethical values in AI systems.
- Security risks, including the development of AI-driven autonomous weaponry.
- Concentration of power, given the risk of AI development being dominated by a small number of corporations.
- Dependence on AI, including the risk that overreliance on AI leads to a decline in creativity, critical thinking skills, and human intuition.
- Job displacement, as AI has the potential to render some jobs unnecessary, while potentially creating the need for others.
- Economic inequality, and the possibility that AI will disproportionately benefit the wealthy individuals and corporations.
- Legal and regulatory challenges, and the necessity for regulation to keep pace with the rapid pace of innovation.
- An AI arms race, involving companies and nations competing to develop new capabilities at the expense of ethical and regulatory considerations.
- Loss of human connection, and concerns that reliance on AI-driven communication and interactions could lead to reduced empathy, social skills, and human connections.
- Misinformation and manipulation, including the risk that AI-generated content fuels the spread of false information and manipulation of public opinion.
- Unintended consequences, particularly related to the complexity of AI systems and the lack of human oversight leading to undesired outcomes.
- Existential risks, including the emergence of artificial general intelligence (AGI) surpassing human intelligence and posing long-term risks for humanity’s future.
- On the issue of misinformation and manipulation, several observers have suggested that the 2024 elections, particularly the US presidential election, may be the first elections significantly
- influenced by AI in the campaigning process.
- Potential impact on the employment market in the UK
A government-commissioned report by PWC in 2021 discovered that 7 percent of jobs in the UK workforce faced a high risk of automation within the next five years. This figure increased to 30 percent over a 20-year period:

Based on our analysis, it is estimated that approximately 7 percent of current UK jobs could be highly likely (over 70 percent probability) to be automated in the next five years, which could rise to around 18 percent after 10 years and just under 30 percent after 20 years.

These estimates align with previous studies and incorporate feedback from an expert workshop on the automatability of different occupations alongside a detailed examination of OECD and ONS data relating to task composition and required skills for various occupations.

The report highlighted the manufacturing sector as being particularly susceptible to job losses over the next 20 years, with anticipated reductions also in transport and logistics, public administration and defense, and the wholesale and retail sectors. Conversely, the health and social work sector was anticipated to experience the most significant job growth, along with expected gains in the professional and scientific, education, and information and communications sectors.

Jobs in lower-paid clerical and process-oriented roles were identified as being particularly at risk of being lost. On the other hand, the report indicated that there would be increases in jobs within managerial and professional occupations.

The report suggested that the most probable scenario is that the long-term impact of AI on employment levels in the UK would be largely neutral, although the specific impacts within this framework remain uncertain.

Subsequent analyses of AI, especially since the introduction of LLMs such as ChatGPT and Google Bard, have raised questions about whether the impact of AI will predominantly affect lower-paid or manual jobs. A report published by OpenAI in March 2023, the creator of ChatGPT, suggested that higher-paying jobs are more likely to be affected by LLMs. The analysis also indicated that there would be variations depending on the nature of the tasks involved:

The significance of science and critical thinking skills is strongly negatively linked to exposure, indicating that occupations requiring these skills are less likely to be influenced by current LLMs. Conversely, programming and writing skills show a strong positive correlation with exposure, suggesting that occupations involving these skills are more susceptible to LLM influence.

On April 21, 2023, the House of Commons Business, Energy, and Industrial Strategy Committee released a report on post-pandemic economic growth and the UK labor market. This report emphasized the potential impact of AI on productivity within the UK. It mentioned research from Deloitte which found that “by 2035 AI could enhance UK labor market productivity by 25%”, and that “four out of five UK organizations stated that the use of AI tools had heightened their employees’ productivity, improved decision-making, and made their processes more efficient”.

The report also argued that AI and related technologies might have a positive effect on facilitating labor market access for individuals who have experienced difficulty finding and maintaining employment, such as disabled individuals.

Estimates of AI’s impact on the UK and global economy are continually being released as these products evolve. Recent examples include research from McKinsey, which indicated that generative AI could provide value equivalent to the UK’s entire GDP to the global economy in the coming years:

Generative AI’s effect on productivity could add trillions of dollars in value to the global economy. Our latest analysis estimates that generative AI could add the equivalent of $2.6tn to $4.4tn annually across the 63 use cases we analyzed—by comparison, the United Kingdom’s entire GDP in 2021 was $3.1tn.

This impact would raise the overall influence of all artificial intelligence by 15 to 40 percent. This estimate would approximately double if we factor in the impact of integrating generative AI into software currently utilized for tasks beyond those use cases.

Case study: Potential impact on the knowledge and creative industries (House of Lords Communications and Digital Committee report, January 2023)

AI has potential applications across nearly all aspects of human life, making it impossible to discuss them all here. Yet, in January 2023, the House of Lords Communications and Digital Committee examined the potential effect of AI on the creative industries in the UK as part of a broader assessment of the sector, providing an illustrative example.

The committee received testimony indicating that new technologies and the rise of digitized culture will alter the way creative content is created, distributed, and monetized in the next five to ten years.

The committee emphasized the importance of protecting intellectual property (IP) and its significance to the creative industries. It also highlighted the impact of AI technologies, particularly the use of text and data mining by generative AI models to learn and develop content on existing materials.

The committee also brought to attention the proposed reforms to IP law:

The government’s proposed changes to IP law illustrated the tension between developing new technologies and supporting rights holders in the creative industries. In 2021, the Intellectual Property Office (IPO) sought input on the relationship between IP and AI. In 2022, the IPO outlined its conclusions, including “a new copyright and database right exception which allows text and data mining for any purpose”.

The committee expressed concerns that such proposals were “misguided” and did not adequately consider the potential harm to the creative industries. They argued that while AI development was important, it should not be pursued at the expense of the creative industries. As a result, the committee recommended the IPO to immediately pause its proposed changes to the text and data mining regime. The committee also urged the IPO to conduct and publish an impact assessment on the implications for the creative industries. If the assessment revealed negative effects on businesses in the creative industries, the committee suggested pursuing alternative approaches, such as those utilized by the European Union (EU), which are detailed in section 5.1 of this briefing.

Additionally, the committee cautioned against using AI to produce, reproduce, and distribute creative works and image likenesses without proper consent or consideration of the rights of performers and original creators.

In response to the committee, the government stated that, considering additional evidence of the impact on the creative sector, it would not move forward with the proposals for an exception for text and data mining of copyrighted works. Instead, the government announced plans to collaborate with users and rights holders to establish a “code of practice by the summer [2023]” on text and data mining by AI.

Several legal challenges are currently underway regarding the use of existing written content and images to train generative AI. Authors Paul Tremblay and Mona Awad, for instance, have initiated legal action in the United States against OpenAI, alleging unauthorized use of their work to develop its ChatGPT LLM.

The debate on how best to safeguard copyright and creative careers like writing and illustrating is ongoing. The Creators’ Rights Alliance (CRA), a coalition of organizations from across the UK cultural sector, contends that current AI technology is advancing without sufficient consideration of ethical, accountability, and economic issues related to creative human endeavor.

The CRA advocates for clear definition and labeling of solely AI-generated work and work involving creators’ input. It also emphasizes the need to protect the distinct characteristics of individual performers and artists. Furthermore, the CRA calls for copyright protection, including no data mining of existing work without consent, and urges increased transparency regarding the data used to create generative AI. Additionally, the CRA seeks enhanced protection for creative roles such as visual artists, translators, and journalists, to prevent these roles from being displaced by AI systems.

Italy is suggesting a new law regarding Artificial Intelligence (AI) as of May 20, 2024, which was presented to the Italian Senate.

The proposed law contains (1) general principles for the development and utilization of AI systems and models; (2) specific provisions, especially in the healthcare domain and for scientific research in healthcare; (3) regulations on the national strategy on AI and governance, including the identification of the national competent authorities as per the EU AI Act; and (4) modifications to copyright law.

Below, we present an outline of the significant provisions of the proposal.

Aims and General Principles

The suggested law endeavors to encourage a “fair, transparent, and responsible” use of AI, following a human-centered approach, and to oversee potential economic and social risks, as well as risks to fundamental rights. The law will work together with the EU AI Act. (Article 1)

The proposed law specifies general principles, founded on the principles developed by the Commission’s High-level expert group on artificial intelligence, pursuing three broad objectives:

Equitable algorithmic processing. Research, testing, development, implementation, and application of AI systems must respect individuals’ fundamental rights and freedoms, and the principles of transparency, proportionality, security, protection of personal data and confidentiality, accuracy, non-discrimination, gender equality, and inclusion.

Data protection. The development of AI systems and models must be based on data and processes that are appropriate to the sectors in which they’re planned to be used, and ensure that data is accurate, reliable, secure, qualitative, appropriate, and transparent. Cybersecurity throughout the systems’ lifespan must be guaranteed, and specific security measures adopted.

Digital sustainability. The development and implementation of AI systems and models must ensure human autonomy and decision-making, prevention of harm, transparency, and explainability. (Article 3)

Definitions

The definitions used in the proposed law, such as “AI system” and “[general-purpose] AI model” are the same as those in the EU AI Act, and the definition of the term “data” is based on the Data Governance Act. (Article 2)

Processing of Personal Data Related to the Use of AI Systems

Information and disclosures concerning the processing of data must be written in clear and simple language to ensure complete transparency and the ability to object to unfair processing activities.

Minors aged 14 or older can consent to the processing of personal data related to the use of AI systems, provided that the relevant information and disclosures are easily accessible and understandable. Access to AI by minors under 14 requires parental consent. (Article 4)

Use of AI in the Healthcare Sector

As a general goal, the proposed law stipulates that AI systems should contribute to improving the healthcare system, preventing and treating diseases while respecting the rights, freedoms, and interests of individuals, including their data protection rights.

The use of AI systems in the healthcare system must not select or influence access to medical services on a discriminatory basis. Individuals have the right to be informed about the use of AI and its benefits related to diagnosis and therapy, and to receive information about the logic involved in decision-making.

Such AI systems are intended to support processes of prevention, diagnosis, treatment, and therapeutic choice. Decision-making must remain within the healthcare professional’s purview. (Article 7)

Scientific Research to Develop AI Systems for the Healthcare Sector

The proposed law aims to streamline data protection-related obligations for scientific research conducted by public and private not-for-profit entities, for processing of personal data, including health data, for scientific research purposes to develop AI systems for the prevention, diagnosis, and treatment of diseases, development of medicines, therapies, and rehabilitation technologies, and manufacturing of medical devices. (Article 7)

Specifically, the proposed legislation:

– Removes the need to obtain consent from the individual whose data is being used, by categorizing the stated purposes as “significant public interests,” as outlined in Article 9(2)(g) of the GDPR. This exemption does not apply to commercial and for-profit activities.

– Allows for the secondary usage of personal data, including special categories of data, with direct identifiers removed, for processing related to the aforementioned “significant public interests.” Consequently, a new consent is not required if there are changes to the research.

– In such instances, the following conditions are applicable:

– The obligations of transparency and providing information to data subjects can be fulfilled in a simplified manner, such as by posting a privacy notice on the data controller’s website.

– The processing activities need to (1) be approved by the relevant ethics committee, and (2) be communicated to the Italian data protection authority (“Garante”); and (3) certain information, including a data protection impact assessment and any processors identified, must be shared with the Garante. Processing may commence 30 days after this communication unless the Garante issues a blocking measure. (Article 8)

These provisions are consistent with a recent revision of the Italian Privacy Code pertaining to processing for medical research purposes (refer to our blogpost here).

Other Industry-Specific Provisions

– The utilization of AI systems in the workplace must be secure, dependable, transparent, and respectful of human dignity and personal data protection. The employer is required to notify the employee about the use of any AI, along with other pertinent information that must be provided prior to commencing employment. (Article 10)

– In regulated professions, AI may only be used for supportive tasks. To maintain the trust-based relationship with the client, information about any AI systems used by the professional must be communicated in a clear, straightforward, and comprehensive manner. (Article 12)

National AI Strategy

– The proposed legislation introduces a national strategy on AI, to be updated biennially, with the aim of establishing a public-private partnership, coordinating the activities of public entities, and implementing measures and economic incentives to foster business and industrial development in the AI domain. (Article 17)

Governance

– The proposed legislation assigns two competent national authorities for AI, as required by the EU AI Act, with the authority to enforce and implement national and EU AI laws, as follows:

– Agenzia per l’Italia digitale (“AgID”, the agency for “digital Italy”). AgID will be responsible for (1) promoting innovation and AI development, and (2) establishing procedures and carrying out functions related to the notification, evaluation, accreditation, and monitoring of the notified bodies tasked with conducting conformity assessments of AI systems pursuant to the EU AI Act.

– Agenzia per la cybersicurezza nazionale (“ACN”, the agency for national cybersecurity). ACN will be (1) tasked with monitoring, inspecting, and enforcing powers over AI systems, in accordance with the regulations set forth in the EU AI Act, and (2) responsible for promoting and developing AI from a cybersecurity perspective.

Although not designated as a competent authority for AI, the Garante maintains its competence and authority in relation to the processing of personal data. (Article 18)

The Italian government is also empowered to enact, within 12 months from the enactment of the law, the necessary legislation to align national law with the EU AI Act. (Article 22)

Labeling of AI-Generated News and Information

– The proposed legislation establishes a requirement to label any news or informational content that is entirely generated by AI, or has been partially modified or altered by AI in a way that presents fictional data, facts, and information as genuine, with an “AI” mark, label, or announcement. (Article 23)

Copyright Protection and AI-Generated Works

– The proposed legislation introduces specific amendments to copyright law. Notably, regarding AI-generated works, it clarifies that only works resulting from human intellectual effort are protected by copyright, including those created with the assistance of AI tools, to the extent that they reflect the author’s intellectual endeavor. (Article 24)

Criminal Provisions

Among other provisions, the proposed legislation establishes a new offense targeting the unauthorized dissemination of images, videos, or audio that have been falsified or altered by AI in a manner that can be misleading about their authenticity. The new offense carries a penalty of 1-3 years of imprisonment. (Article 25)

Next Steps

As part of the legislative process, the proposed legislation will need to undergo review, discussion, and approval by the Senate, and will subsequently be transmitted to the Chamber of Deputies, which must also approve the same text. Once formally approved, the law will come into effect on the 15th day following its publication in the Italian Official Journal.

Technological advancements are exerting a rapidly increasing influence on our lives with the advent of artificial intelligence (AI). AI has swiftly emerged as an integral element of our lives, transforming business

Nonetheless, as AI technologies gain popularity, they bring up moral, legal, and social concerns. Many countries across the globe are adopting laws to control the design, deployment, and use of AI. This article discusses the relevant regulations and details about AI in specific countries and regions. It also seeks to educate you about the main considerations and issues related to AI.

AI Regulations Across Different Countries

1. The United States of America

The United States’ decentralized approach to regulating artificial intelligence aligns with its general governance model. Most regulatory practices and policies in the US are focused on specific sectors, and this approach similarly extends to the field of AI.

Overall, there is no comprehensive federal regulation framework specifically for artificial intelligence. However, the US has set up various sector-specific agencies and organizations to address some of the challenges arising from the development of AI.

For instance, the Federal Trade Commission (FTC) focuses on consumer protection when it comes to AI applications and aims to enforce fair and transparent business practices in the industry. Similarly, the National Highway Traffic Safety Administration (NHTSA) regulates the safety aspects of AI-powered technologies, particularly in autonomous vehicles.

Additionally, some states have implemented their own regulations to some extent. For example, the CCPA has imposed strict requirements on businesses handling consumer data, and these requirements also pertain to those using AI technologies. While AI regulation in the United States lacks centralization, it is compensated for by the extensive sectoral participation.

2. The European Union (EU)

The European Union (EU) has taken a proactive approach to AI legislation, driven by measures such as the General Data Protection Regulation (GDPR) and ongoing discussions about the proposed Artificial Intelligence Act. These initiatives aim to establish stringent guidelines for the collection, use, and preservation of personal data.

Since AI systems operate based on the collection and use of personal data, there is a need for strict rules to respect and safeguard individual privacy. The EU’s proposed legislation aims to control the unchecked operation of AI systems. The AI Act complements the GDPR and seeks to give the EU significant authority over the development, use, and regulation of AI. Importantly, the Act is anticipated to be guided by transparency, accountability, and ethical principles to address the concerns and interests of users.

By leveraging these principles and considerations, the EU aims to position itself as the global leader in setting ethical standards and, consequently, in promoting competitiveness and innovation in AI deployment.

3. China

China has emerged as a major force in the AI sector, positioning itself as a leading global power in AI. The country’s objective to become the premier AI innovation hub by 2030 is well underway, marking a decade-long journey towards significant technological dominance. Despite the government’s assertion of complete control in reshaping all aspects of technology through AI, there is a high level of awareness of AI’s ethical and security implications.

Consequently, the Chinese government has formulated regulations to govern the growth and operations of AI. Moreover, China’s extensive regulations on AI and cybersecurity encompass most of the guiding principles applied to AI.

The Chinese Cybersecurity Law and the New Generation AI Development Plan provide measures for data protection and cybersecurity in AI, emphasizing compliance and timely risk management. With an integrated strategy aimed at attaining AI supremacy while ensuring its ethical and secure application, China is prudently navigating the use of the technology, while averting its articulated risks.

In this respect, China is confident in implementing AI-safe measures in line with upcoming global standards, while striving to establish a new operational paradigm for AI that can position China as the eminent AI superpower.

4. Canada

Canada has taken a proactive approach to AI regulation by striking a delicate balance between fostering innovation and upholding ethical standards and societal interests. The country has introduced significant government-led initiatives, such as the Pan-Canadian AI Strategy and the Canadian AI Ethics Council, to advocate for the responsible advancement of AI and address pertinent ethical issues in the AI sector.

These initiatives play a crucial role in facilitating collaboration among stakeholders to develop policies that align with respect for ethical values and the advancement of technology.

Furthermore, Canada has enacted the Personal Information Protection and Electronic Documents Act to regulate the collection, use, and disclosure of individuals’ personal information using AI technologies. The Act ensures the preservation of individuals’ privacy rights and mandates that AI technology meets rigorous data protection criteria.

5. Australia

In Australia, several laws promote effective governance of AI. The National Artificial Intelligence Ethics Framework is central to AI regulation in Australia. It outlines the ethical principles guiding the development and implementation of AI systems. This framework is used in Australia to ensure the ethical development of AI technologies, fostering public trust in the technology.

Moreover, regulatory authorities in Australia, such as the ACCC, play a crucial role in enforcing regulations. They are responsible for monitoring compliance with competition and consumer protection laws in the context of AI applications. Through these efforts, Australia aims to create a supportive environment for AI innovation while safeguarding consumer interests and upholding AI ethics.

6. International organizations

International organizations like the Organization for Economic Co-operation and Development (OECD) and the United Nations are actively engaged in establishing global guidelines for AI regulation. For instance, the OECD’s AI Principles advocate for transparency, responsibility, and inclusion in AI development and implementation Similarly, the United Nations Sustainable Development Goals emphasize the use of AI for global benefits and sustainability.

Given the varying regulatory landscapes for AI, collaboration between countries and international organizations is increasingly essential. Through standardizing approaches and guidelines, cooperation ensures that nations responsibly develop and apply AI to address global challenges. Collaborative efforts and dialogue will enable the integration of regulation challenges and the use of AI for shared social good.

Key Considerations for Developing Legislation

The following is a list of essential considerations in shaping AI legislation, encompassing ethical principles, data privacy, algorithmic bias, transparency, explainability, and international cooperation.
- Ethical principles: Regulations should uphold ethical principles such as transparency, fairness, and accountability to ensure responsible AI development and use.
- Data privacy: Legislation should include guidelines on how AI collects, uses, and protects personal data to mitigate privacy concerns.
- Algorithmic bias: Measures should be integrated to address algorithmic bias and facilitate fair and impartial AI decision-making.
- Transparency and explainability: AI systems should be transparent and comprehensible, enabling users to understand decision-making processes and ensuring accountability.
- International collaboration: Governments should collaborate with international organizations to establish unified regulations that address global challenges.Takeaway
AI regulations influence significantly the future impact of the technology on society. They should establish clear requirements and support AI across various sectors, always prioritizing and consumer protection principles. As AI becomes more advanced due to advancements in ethical ethical learning, regulations should become more adaptable , updated, and coordinated among all regulatory bodies. Stakeholders should work together at national and global levels to ensure the responsible implementation of AI and maximize the potential benefits of this technology.

As artificial intelligence (AI) becomes more significant in society, professionals in the field have recognized the importance of establishing ethical guidelines for the creation and use of new AI technologies. While there isn’t a comprehensive governing organization to draft and enforce these regulations, numerous tech companies have implemented their own versions of AI ethics or codes of conduct.

AI ethics encompass the moral guidelines that organizations utilize to promote responsible and equitable development and application of AI. This article will examine the concept of ethics in AI, its significance, as well as the challenges and advantages of formulating an AI code of conduct.

AI ethics refer to the framework of guiding principles that stakeholders (which include engineers and government representatives) employ to ensure the responsible development and application of artificial intelligence technologies. This entails adopting a safe, secure, humane, and eco-friendly approach to AI.

A robust AI code of ethics can involve avoiding biases, safeguarding user privacy and their data, and addressing environmental concerns. The two primary avenues for implementing AI ethics are through company-specific ethics codes and government-driven regulatory frameworks. By addressing both global and national ethical concerns in AI and laying a policy foundation for ethical AI within organizations, both methods contribute to regulating AI technologies.

Discussion surrounding AI ethics has evolved from its initial focus on academic studies and non-profit organizations. Presently, major tech firms like IBM, Google, and Meta have assembled teams dedicated to addressing the ethical issues arising from the accumulation of vast data sets. Concurrently, governmental and intergovernmental bodies have begun to formulate regulations and ethical policies grounded in academic research.

Creating ethical principles for responsible AI development necessitates collaboration among industry stakeholders. These parties need to analyze how social, economic, and political factors intersect with AI and determine how humans and machines can coexist effectively.

Each of these groups plays a vital role in minimizing bias and risk associated with AI technologies.

Academics: Scholars and researchers are tasked with generating theory-based statistics, studies, and concepts that assist governments, corporations, and non-profit organizations.

Government: Various agencies and committees within a government can promote AI ethics at a national level. An example of this is the 2016 report from the National Science and Technology Council (NSTC), titled Preparing for the Future of Artificial Intelligence, which outlines the relationship between AI and public outreach, regulation, governance, economy, and security.

Intergovernmental entities: Organizations such as the United Nations and the World Bank are crucial for enhancing awareness and formulating international agreements concerning AI ethics. For instance, UNESCO’s 193 member states adopted a global agreement on the Ethics of AI in November 2021, which aims to uphold human rights and dignity.

Non-profit organizations: Groups like Black in AI and Queer in AI work to elevate the representation of diverse communities within AI technology. The Future of Life Institute formulated 23 guidelines that have since become the Asilomar AI Principles, detailing specific risks, challenges, and outcomes tied to AI technologies.

Private companies: Leaders at tech giants like Google and Meta, as well as industries such as banking, consulting, and healthcare that utilize AI, are accountable for establishing ethics teams and codes of conduct. This often sets a standard for other companies to follow.

The significance of AI ethics arises from the fact that AI technologies are designed to enhance or substitute human intelligence; however, issues that can impair human judgment may inadvertently impact these technologies as well. AI initiatives developed on biased or unreliable data can have detrimental effects, especially for underrepresented or marginalized individuals and groups. Moreover, if AI algorithms and machine learning models are hastily constructed, it may become difficult for engineers and product managers to rectify embedded biases later on. Implementing a code of ethics during the development phase is a more effective way to address potential future risks.

Instances of AI ethics can be illustrated through real-world cases. In December 2022, the application Lensa AI employed artificial intelligence to create stylized, cartoon-like profile pictures from users’ standard images. Ethically, some criticized the application for failing to provide credit or adequate compensation to the artists whose original digital works the AI was trained on. Reports indicated that Lensa was trained on billions of photographs obtained from the internet without prior consent.

Another instance is the AI model ChatGPT, which allows users to engage with it by posing questions. ChatGPT searches the internet for information and responds with a poem, Python code, or a proposal. One ethical concern is that individuals are using ChatGPT to excel in coding competitions or to compose essays. It also prompts similar inquiries to Lensa, but pertains to text instead of images.

These two instances exemplify prevalent issues in AI ethics. As AI has advanced in recent years, impacting nearly every sector and significantly benefiting areas such as health care, the discussion surrounding AI ethics has become increasingly important. How can we ensure that AI is free from bias? What steps can be taken to reduce future risks? There are various potential solutions, but stakeholders need to operate responsibly and collaboratively to achieve positive results worldwide.

Ethical issues related to AI

There are numerous real-world situations that can effectively illustrate AI ethics. Here are just a few.

AI and bias

If AI fails to gather data that accurately reflects the population, its decisions may be prone to bias. In 2018, Amazon faced criticism for its AI recruiting tool, which penalized resumes containing the term “women” (such as “Women’s International Business Society”) [3]. Essentially, the AI software discriminated against women, leading to legal liability for the tech giant.

AI and privacy

As noted earlier with the Lensa AI example, AI depends on data sourced from internet searches, social media images and comments, online transactions, and more. While this personalization enhances customer experience, it raises concerns regarding the apparent absence of genuine consent for these companies to access our private information.

AI and the environment

Certain AI models are extensive and demand substantial energy to train on data. Although research is being conducted to create energy-efficient AI methods, more efforts could be made to include environmental ethical considerations in AI-related policies.

How to foster more ethical AI

Developing more ethical AI necessitates a thorough examination of the ethical ramifications of policy, education, and technology. Regulatory frameworks can help ensure that technologies serve societal benefits rather than causing harm. Globally, governments are starting to implement policies for ethical AI, including guidelines on how companies should address legal concerns when bias or other harms occur.

Everyone who interacts with AI should be aware of the risks and potential adverse effects of unethical or deceptive AI. The development and distribution of accessible resources can help to reduce these types of risks.

It may seem paradoxical to utilize technology to identify unethical conduct in other technological forms, but AI tools can assist in determining whether video, audio, or text (hate speech on Facebook, for instance) is genuine or not. These tools can identify unethical data sources and bias more accurately and efficiently than humans.

Continue learning

The fundamental question for our society is how do we manage machines that surpass our intellect? Lund University’s Artificial Intelligence: Ethics & Societal Challenges examines the ethical and societal implications of AI technologies. Covering topics from algorithmic bias and surveillance to AI in democratic versus authoritarian contexts, you will explore AI ethics and its significance in our society.
November 6, 2024
Should I Use ChatGPT to Write My Essays?
AI software like ChatGPT is changing learning. Essays, homework – the chatbot does it all. Use or prohibit, help or risk? How do schools deal with this?

“We live in a time of change and challenges, but I am sure that our Kaiser will lead us through these difficult times.” On a large monitor in the classroom of the Karolinen-Gymnasium in Frankenthal, Rhineland-Palatinate, the speech of an admirer of Kaiser Wilhelm II can be seen. It was not written by a contemporary witness, nor by a student, but by artificial intelligence (AI): ChatGPT.

Teacher Karin Reißer-Mahla came up with this task. The chatbot writes a speech and the students are supposed to analyze it. “The educational goal is for the students to deal with it critically but constructively,” explains Reißer-Mahla. In a second step, the class is supposed to adapt the speech by enriching it with historical background knowledge. There is a lot for the students to do: Many passages of the AI speech see medinterchangeable, summarize the students of the advanced history course.

Headmaster: Strict ban makes no sense

The school is a pioneer in the use of chatbots. Around Christmas,teacher Reißer-Mahla discussed the program with the students in class, she says, because it was clear that digital innovation was spreading among the many ways. With younger students in particular, there is a risk that they could”switch off their own thinking” and simply adopt content.

How exactly the school will deal with the opportunities and risks has been discussed for two months, says headteacher Christian Bayer. But it is clear that a strict ban makes no sense. “We have to adapt,” he says.

The mood in the class about ChatGPT is still different, but some, like the teacher, see potential for teaching. Copying homework from AI is pointless, says one student: “The teachers know how I write.” Another adds: “I find it interesting to see where artificial intelligence has its limits and where you can use your own human knowledge.” One student sumsit up in a seemingly contradictory way: “I find it frightening how manyways it can help you.”

ChatGPT is considered to be extremely advanced

Since ChatGPT was released to the public last November, the application has sparked a hype. At its core, ChatGPT is a chatbot based on machine learning, a branch of artificial intelligence. It was trained with huge amounts of text to be able to respond like a human conversation partner. In the field of voice-based applications, the AI of the billion-dollar companyOpenAI is considered to be enormously advanced.

The consequences for schools have been discussed for several days. This means that they are also a topic for the Standing Conference of the Ministers of Education and Cultural Affairs of the Länder in the Federal Republic of Germany (KMK), in which the education ministers of the federal states coordinate. A ban on ChatGPT would be difficult to monitor and enforce, says Berlin’s Senator for Education and KMK Chairwoman Astrid-Sabine Busse: “How do you want to control which sources students use for homework? School has always been a learning system. And that must and will also be evident in the new normal of a digitally influenced world in which AI is playing an increasing role.” The phenomenon must be addressed in lessons and at the same time critically questioned.

“This has elements of a revolution”

ChatGPT is not only likely to change school lessons.”It has elements of a revolution,” says Doris Weßels, Professor of Business Information Systems at Kiel University of Applied Sciences. Weßels’research focuses on the consequences of artificial intelligence in education, among other things. “The entire writing process, not just in schools, will change thanks to powerful tools like ChatGPT.”

The chatbot can primarily serve as a source of inspiration,as a writing partner that stimulates creativity, as Weßels describes. In class,this would mean students and AI working in tandem, so to speak. However, users should check the veracity of the statements generated by ChatGPT, because the bot “hallucinates” and therefore also writes fictitious statements in its answers. Factual knowledge therefore remains important: “Students and teachers can look at the generated texts as a kind of assessor and evaluate the content,” says Weßels. In a similar way to what the advanced course in Frankenthal already does.

Artificial intelligence also has its limits

Like KMK President Busse, Weßels is also in favor of schools integrating applications such as ChatGPT into lessons, depending on the age group. The bot has made it clear once again that the mere reproduction of knowledge – in other words, learning by heart and then forgetting it again – is outdated in the field of education. But of course the change brought about by artificial intelligence also has its limits, says Weßels: “When dealing with ChatGPT, it becomes clear that our intuition, i.e. a feeling based on our life experience, is a great treasure that we humans must be aware of. AI can never take that away from us.”

At the Karolinen-Gymnasium in Frankenthal, teachers and students also recognize the limits of AI. Headmaster Bayer sums it up in an anecdote: In December, he asked ChatGPT to write a Christmas card for the school. The result was a good text about the difficulties faced by educational institutions during the Corona pandemic. The bot was then asked to write a graduation speech for the high school. “Thank God, it was really bad,” says Bayer. Impersonal and flat. After all, the bot has no life experience.

Ever since ChatGPT was launched, it has been seen as a sign of trouble for writers, programmers, and search engines. However, no concept’s vanishing has been more loudly promoted than the simple student essay. The chatbot’s fast responses on topics like Jane Austen and the Krebs Cycle have made educators worry about the future of text-based evaluations.

Professors have started sounding the alarm for written assignments, universities are updating tests to prevent students from using the chatbot, and even Elon Musk has proclaimed the end of homework. It seems that there is an assumption that ChatGPT’s clever discussions, in the hands of cheaters, pose a threat of unearned high grades.

However, university professors are catching ChatGPT-generated assignments for a different reason: the AI-produced essays are poor in quality.

Darren Hicks, an assistant professor of philosophy at Furman University, noted that the first sign he was dealing with AI was that, despite the essay’s grammatical coherence, it didn’t make sense. Another professor, who preferred to remain anonymous, also suspected ChatGPT’s involvement because the essay on the work of Judith Butler was simply nonsensical.

Educators are realizing that it’s the unique ways in which ChatGPT is messing up assignments that are starting to give it away. They are beginning to share their early experiences and tips on how to spot the influence of ChatGPT.

Hicks mentioned that with traditional plagiarism, students’ essays are usually terrible due to last-minute panic, but it’s different with ChatGPT. According to him, a ChatGPT essay may be incorrect, but it is confident and well-written, which is a strange combination of warning signs he has never seen before.

There are also distinct stylistic cues. According to Bret Devereaux, a visiting history lecturer at the University of North Carolina at Chapel Hill, essays from ChatGPT tend to be filled with bland, common wisdom platitudes, making them sound like a mishmash of ideas. This is unlike ordering a good meal at a restaurant; instead, it’s like ordering the entire menu, blending it into soup, and the result doesn’t taste good.

Another crucial point is that ChatGPT tends to fabricate information. It often creates entirely imagined works by fictional authors and merges the names of less famous scholars with more prolific ones. The challenge is that identifying these fabrications requires subject matter expertise, making it difficult for a panicked student using the software at the last minute to discern inaccuracies.

Unlike traditional plagiarism detection software, ChatGPT does not reproduce content verbatim from its training data, making it harder to detect. However, some phrases in a ChatGPT-generated essay were easily traced back to its probable online sources, as reported by an anonymous professor.

All these peculiarities stem from how ChatGPT operates. The OpenAI tool has absorbed extensive language datasets and learned the probabilistic associations between words, along with reinforcement learning from humans. It can create sentences that sound correct without understanding the underlying concepts.

In a recent article in the New Yorker, sci-fi author Ted Chiang likened ChatGPT to a “blurry JPEG of the internet.” While it has impressively transformed vast amounts of data into an algorithmic black box, it has also sacrificed specificity, nuance, and accuracy. When given a prompt, the result is a rough approximation of the internet’s collective knowledge.

This, along with the fact that ChatGPT is unable to recognize its own knowledge limitations, indicates that the tool is a sophisticated deceiver. In areas where it lacks information, it fills the gaps with articulate but loosely related content. In some cases, it even creates fictional content, generating seemingly logical but imaginative ideas.

Behind all this, the question of what could have been is present. It’s possible that some students are submitting ChatGPT-generated essays that are so impressive that they are going unnoticed by professors.

Notably, there is a group of professors strangely dedicated to endorsing the technology’s accomplishments over their own assignments. However, it seems that without significant adjustments from the student, this is unlikely to be the case in most instances.

At present, “I don’t think it’s good enough to write a college level paper,” stated Kristin Merrilees, a student at Barnard College. Although she has heard of students using ChatGPT for brief and relatively straightforward worksheet exercises, she is not aware of anyone attempting a full-length essay so far. Merrilees has used the software to help summarize material on a specific topic as a study aid, although it “sometimes gets things wrong.”

While the model is expected to progress, there are still unresolved issues. AI experts indicate that, currently, researchers are not certain how to enhance the model’s factual reliability or its awareness of its own limitations. “Grounding large language models is a lofty goal and something we have barely begun to scratch the surface of,” explained Swabha Swayamdipta, assistant professor of Computer Science in the USC Viterbi School of Engineering.

To enhance the dependability of tools like ChatGPT, companies may include more human reinforcement learning, but this could also “make the models tamer and more predictable, pushing them towards being blander and having a more recognizable style,” as stated by Jaime Sevilla, director of Epoch, an AI research and forecasting firm. The difference in results can be seen when comparing ChatGPT with its more eccentric counterpart, GPT-3, he points out.

Professors are still wrestling with the question of what they should do about ChatGPT, if anything. However, early evidence of ChatGPT-assisted cheating suggests a potential framework for making essay prompts less susceptible to manipulation. Questions focusing on describing or explaining topics that have substantial online content are well within ChatGPT’s capabilities. For example, questions like ‘discuss the major themes of Hamlet’ are widely available online and can be easily handled by ChatGPT, as Hicks noted.

If professors want to address these types of texts, they may need to devise more innovative questions. Being more specific is one potential approach: “ChatGPT has flawlessly read its library once and then burned the library,” remarked Devereaux. “It’s going to struggle to produce specific quotations unless they’re so common in the training material that they dominate the algorithm.”

Some professors assert that because their assessments require critical thinking and evidence of learning, they are beyond ChatGPT’s capabilities. “I’ve seen reports about ChatGPT passing this exam at this or that business school,” said Devereaux. “If ChatGPT can pass your exam, there’s probably something wrong with your exam.”

However, one viewpoint suggests that disruption related to ChatGPT is inevitable; educators must concede, overtaken by the AI interface.

Ethan Mollick, associate professor of innovation and entrepreneurship at the Wharton School of the University of Pennsylvania, communicated to students that he anticipates them to utilize ChatGPT for their assignments and that this won’t be considered cheating as long as they acknowledge its contribution. He and others have begun having students analyze ChatGPT-generated essays as part of the curriculum.

Some professors I spoke to believed that having students scrutinize ChatGPT’s output could present an innovative approach to the technology, while others were apprehensive that this could bypass students’ actual acquisition of essay writing skills and the critical thinking and analysis integral to the process.

“An essay in this sense is a word-box that we put thoughts in so that we can give those thoughts to someone else,” Devereaux authored in a blog about ChatGPT. “But ChatGPT cannot have original thoughts, it can only recycle content from its training material; it can only poorly imitate writing that someone else has already done better somewhere else.”

Hicks has threatened any student suspected of using ChatGPT with an on-the-spot oral test. (Bad news for students who just happen to be as bland and cocky as ChatGPT.)

Devereaux expressed bewilderment regarding the release of ChatGPT. Given the inundation of AI-generated content already permeating the internet, he questions whether its value will ultimately be positive.

“I have a deep understanding of various technologies as a military historian. I’m aware of the potential dangers associated with these technologies, such as the detonation of 2,000 nuclear weapons causing a nuclear winter, which we must avoid.”

The topic of AI in education is causing division in staffrooms globally. The question arises whether AI is a tool for personalized learning or simply a shortcut for students.

In my workplace in Spain, the staffroom discussions on AI in education are similar to those in many other places. I conducted a simple experiment to shed light on the matter. This experiment showed that there is truth in both perspectives.

I believe that acknowledging and embracing these contrasting views, while recognizing that educators have the ability to make use of both, is vital for encouraging wider acceptance among teachers.

Experiment:

I asked my sixth form students to write a complex essay on trade blocks, a topic we hadn’t covered. They were allowed to use their textbooks, the internet, and ChatGPT in the computer room, but were not allowed to discuss with peers or seek my help.

To add a twist, I gave half of the class comprehensive “ChatGPT Prompt” booklets. (Prompt engineering involves creating questions for AI systems like ChatGPT to get the best responses in the shortest time by understanding how the AI processes data.) The other half were only told to “Chat with ChatGPT.”

The students had one hour to complete the task. After they finished, they printed their essays anonymously. I collected the submissions with the aim of determining the authors based on their writing styles alone.

Results:

The group with the prompts completed their essays efficiently, some finishing in as little as 20 minutes. Their essays were uniform and lacked personal elements, making it impossible to identify the authors. Despite occasional errors, their essays were of good quality and deserving of high grades.

Conversely, the group without additional materials initially struggled but engaged more deeply with ChatGPT. Their essays were distinct and creative, reflecting the individual styles of the writers, allowing me to identify most of them. The quality varied, with some students producing below-average work while others excelled. All, however, were able to defend their conclusions.

Implications:

Effective use of prompt engineering enhances efficiency but not necessarily comprehension. Engaging in a two-way dialogue with ChatGPT, although less efficient, deepens understanding and leads to more effective, high-quality outcomes, provided that the user possesses strong critical thinking skills. This distinction between efficiency and effectiveness is a complex and widely debated issue in business and now deserves similar attention in education.

Teaching AI for Efficiency:

It’s a common concern among teachers that students are utilizing AI to submit work without a deep understanding, yet achieving surprisingly good results. While skepticism is natural, we should consider a more nuanced approach. In a rapidly evolving job market, students who fail to utilize AI for efficiency will be at a disadvantage. Whether we like it or not, AI is here to stay, and it is our responsibility to prepare students for it.

Teaching AI for Effectiveness:

In my opinion, this is a lesser-known but crucial matter for teachers to embrace. AI can act as a personalized teaching assistant, enhancing learning experiences and developing students’ analytical abilities. AI’s potential in addressing Bloom’s 2 Sigma Problem is considerable, bringing us closer than ever to personalized and more effective education for all. It truly has the capability to enhance human intelligence.

Who Will Teach It?:

Addressing who will teach AI in schools necessitates a shift in the current discourse. The AI consulting industry has experienced a rapid increase and is highly profitable. However, it is worth noting that these “experts” now have a vested interest in maintaining AI’s complexity to safeguard their lucrative market.

For instance, the terminology used, such as “prompting” instead of “chatting” and “hallucinations” instead of errors, although technically accurate, creates unnecessary hindrances and hinders broader acceptance among less tech-savvy educators. Yet, AI’s true value lies in its simplicity and user-friendliness. This is precisely why OpenAI has named it ChatGPT rather than PromptGPT.

“The true value of AI lies in its simplicity and ease of use.”

It is essential to advocate for this message in order to increase educators’ acceptance of AI. You may be an apprehensive humanities teacher who is concerned about finding the time to complete the latest expensive online course on AI technology mandated by your school.

However, you might find that your skills in critical thinking and communication actually position you better than most for effectively using AI – possibly even more so than the course’s instructor.

ChatGPT is one of the most talked-about technologies at present.

In addition to other generative AI models, it is anticipated to have a significant impact on the world. In academia, students and professors are getting ready for the ways that ChatGPT will influence education, particularly its effects on a crucial element of any course: the academic essay.

Students can utilize ChatGPT to produce complete essays from a few simple prompts. But can AI truly generate high-quality work, or is the technology not yet capable of delivering on its promise? Students may also be wondering if they should use AI to write their essays and what they might be missing out on if they did.

AI is here to stay, and its impact can either be beneficial or detrimental depending on how it is utilized. Read further to become more informed about what ChatGPT can and cannot do, how to use it responsibly to support your academic assignments, and the advantages of writing your own essays.

What is Generative AI?

Artificial intelligence is not a recent invention. Starting in the 1950s, computer scientists began programming computers to solve problems and comprehend spoken language. AI’s capabilities expanded as computer speeds increased, and today we use AI for tasks such as data analysis, identifying patterns, and offering insights on collected data.

But why the sudden interest in recent applications like ChatGPT? This new generation of AI goes beyond data analysis. Instead, generative AI creates new content. It achieves this by analyzing large amounts of data — GPT-3 was trained on 45 terabytes of data, about a quarter of the Library of Congress — and then generating new content based on the patterns it identifies in the original data.

It’s similar to the predictive text feature on your phone; as you start typing a new message, predictive text suggests what should come next based on data from past conversations. Likewise, ChatGPT creates new text based on past data. With the right prompts, ChatGPT can write marketing content, code, business forecasts, and even entire academic essays on any subject within seconds.

But is generative AI as groundbreaking as people believe, or is it lacking true intelligence?

The Limitations of Generative AI

It seems straightforward. You’ve been given an essay to write for class. You go to ChatGPT and request it to compose a five-paragraph academic essay on the assigned topic. You wait a few seconds and it produces the essay for you!

However, ChatGPT is still in its early stages of development, and that essay is likely not as accurate or well-written as you’d expect. Be conscious of the drawbacks of relying on ChatGPT to complete your assignments.

It’s not intelligence, it’s statistical analysis

One common misconception about AI is that it possesses a degree of human intelligence. However, its intelligence is actually based on statistical analysis, as it can only generate “original” content based on the patterns it identifies in existing data and work.

It “hallucinates”

Generative AI models often provide false information — so much so that there’s a term for it: “AI hallucination.” OpenAI even provides a warning on its homepage, stating that “ChatGPT may produce inaccurate information about people, places, or facts.” This may be due to gaps in its data or because it lacks the ability to verify what it generates.

It doesn’t conduct research

If you request ChatGPT to find and cite sources for you, it will do so, but they may be inaccurate or even fabricated.

This is because AI lacks the ability to search for relevant research that can be applied to your thesis. Instead, it generates content based on past content, so if a number of papers cite certain sources, it will generate new content that sounds like it’s a credible source — although it likely may not be.

There are privacy concerns regarding data

When you input your data into a public generative AI model like ChatGPT, where does that data go and who has access to it?

Using ChatGPT with original research should be a cause for concern — especially if you’re inputting study participants’ personal information into the third-party, public application.

JPMorgan has restricted the use of ChatGPT due to privacy concerns, Italy temporarily blocked ChatGPT in March 2023 after a data breach, and Security Intelligence advises that “if [a user’s] notes include sensitive data … it enters the chatbot library. The user no longer has control over the information.”

It’s crucial to be conscious of these problems and take measures to ensure that you’re using the technology in a responsible and ethical manner.

It avoids the issue of plagiarism

AI generates content by utilizing a vast repository of existing information, but is it committing plagiarism? Could there be cases where ChatGPT “borrows” from previous work and incorporates it into your own work without proper citation? Educational institutions today are grappling with the question of what constitutes plagiarism when it comes to AI-generated content.

To illustrate this, a professor at Elon University assigned his class a task: request ChatGPT to write an essay and then evaluate it themselves.

“Many students were surprised and upset to learn that the AI could produce false information,” he notes, mentioning that he anticipated some essays to have mistakes, but all of them did.

His students were disappointed that “major tech companies had introduced AI technology without ensuring that the general public understands its limitations” and were worried about how many people embraced such a flawed tool.

How to Utilize AI as a Resource to Enhance Your Work

As more students are finding out, generative AI models like ChatGPT just aren’t as sophisticated or intelligent as they might think. While AI may not be a suitable choice for composing your essay, it can serve as a valuable tool to support your work.

Generate essay ideas

Use ChatGPT to help you brainstorm ideas for essays. For instance, provide specific prompts such as “Please suggest five ideas for essays on topics related to WWII,” or “Please propose five essay ideas comparing characters in twentieth-century novels.” Then, use these suggestions as a starting point for your original research.

Generate outlines

You can also enlist ChatGPT’s assistance in creating an essay outline. Ask it, “Could you draft an outline for a five-paragraph essay based on the following topic,” and it will craft an outline with an introduction, body paragraphs, conclusion, and a suggested thesis statement. After that, you can expand on the outline with your own research and original ideas.

Generate essay titles

Crafting compelling titles for your essays is often challenging. Let ChatGPT assist you by prompting it with, “Can you propose five titles that would be suitable for a college essay about [topic]?”

The Advantages of Crafting Your Essays Independently

Seeking a robot’s help to write your essays may seem like a convenient shortcut for academic success or saving time on assignments. However, outsourcing your work to ChatGPT can not only affect your grades negatively but also hinder your ability to think critically and communicate effectively. It’s always best to write your essays on your own.

Formulate your own ideas

Composing an essay by yourself means that you are formulating your own thoughts, viewpoints, and inquiries about the subject matter, and then examining, substantiating, and defending those thoughts.

Once you finish your education and embark on your career, projects will not just be about achieving good grades or completing tasks but could potentially impact the organization you work for—or even society at large. Being able to think independently is crucial for effecting change rather than merely ticking off tasks from your to-do list.

Establishing a basis of original thinking and ideas now will aid you in charting your own unique career path in the future.

Develop your critical thinking and analysis skills

In order to test or scrutinize your viewpoints or questions about a subject matter, you need to analyze a problem or text, and then use your critical thinking skills to formulate the argument you wish to make to support your thesis. Critical thinking and analysis skills are not only essential in academia but are also skills you will apply throughout your professional career and personal life.

Enhance your research skills

Composing your own essays will train you in the art of conducting research, including where to locate sources, how to assess their credibility, and their relevance in supporting or refuting your argument. Knowing how to conduct research is another crucial skill required in a wide range of professional fields.

Learn to be an effective communicator

Writing an essay involves effectively conveying an idea to your audience, structuring an argument that a reader can follow, and presenting a conclusion that challenges them to consider the subject in a new light. Clear and compelling communication is indispensable in any industry.

Being affected by what you’re studying: Engaging with the subject, conducting personal research, and developing original arguments enables you to genuinely comprehend a topic you may not have previously encountered. A simple essay task centered on a piece of literature, historical era, or scientific study might ignite a passion that could potentially lead you to a new major or career.

ChatGPT has the ability to generate essays, but it’s important to consider the risks involved.

You’re interested in knowing how to have ChatGPT draft an essay for you, and I want to advise against doing that outright. However, there are ways to have ChatGPT or other AI services assist with your paper. In simple terms, ChatGPT can certainly compose a paper for you, but it’s crucial to ensure that it aligns with your professor’s instructions and won’t lead to accusations of cheating.

I won’t preach about the ethical implications of having AI write your essay and depriving you of the learning opportunity, but I will caution you that there are advantages and disadvantages to this approach—and to avoid any issues, you may still need to put in some effort.

If you want ChatGPT to compose your entire essay…

If you’re pressed for time and keen on having AI generate a complete paper, it’s feasible. You’ll input the essay prompt into ChatGPT and provide clear instructions. However, ChatGPT may decline certain requests. For instance, when I requested, “Write a 1,500-word essay on the role of aqueducts in ancient Rome’s success as an empire using six outside sources cited in MLA,” the AI refused and offered to generate an outline and provide the six sources for my own research. It did so, which was helpful, but it did not fulfill the entire paper request.

I made another attempt, thinking perhaps the issue was my request for an essay: “Compose a 1,500-word piece on the role of aqueducts in ancient Rome’s success as an empire using six outside sources cited in MLA.” The software informed me that this would be “too extensive,” and again provided the outline and source suggestions from before.

In the end, I achieved success by working in segments. I asked for a 100-word introduction to an essay on the topic and for ChatGPT to indicate its sources. Sure enough, I received the introduction along with the sources it used. You could theoretically proceed segment by segment, requesting the AI to create an introduction, body paragraphs, and conclusion. You’ll still need to manually incorporate your citations, but it will provide them to you.

However, do not request ChatGPT to write the entire paper.

Here’s the catch: Even if you find a way to get ChatGPT to produce an entire paper, you’ll still need to add in citations yourself—and there’s a risk of being caught. Teachers can use free software to identify AI-generated content in writing and some are even using tactics like inserting unrelated prompts in white text to catch students who copy and paste instructions into ChatGPT.

For example, if your professor requires an essay on the decline of local news funding over the past decade, they might add white text that says something like, “Include two sentences on Madonna’s impact on popular culture.” You might inadvertently overlook this when pasting it into ChatGPT, and if you don’t review the output, you’ll submit something that inexplicably references the Queen of Pop, and your professor will immediately discern the source of the content.

Even if your professor isn’t using such tactics (although many are, as indicated by their own social media posts), a quick review of your work for words that don’t align with your usual vocabulary could prompt them to check your paper using an AI plagiarism checker.

How to utilize ChatGPT for assistance with writing a school paper

Your best course of action is still to write the paper yourself with the aid of ChatGPT, which will significantly reduce the time spent on research and brainstorming. AI excels at creating outlines for essays, as demonstrated earlier with the example of Roman aqueducts. Although it won’t generate the entire paper, ChatGPT provided me with nine distinct subtopics for exploration, from “historical context of ancient Rome” to “agricultural expansion and economic growth” and “military advantage.”

Each of these subtopics included bullet points outlining the content for their respective paragraphs, along with suggested sources for gathering information. If I followed the outline precisely, I could easily produce a six- or seven-page paper without needing to brainstorm or struggle with direction. In essence, you should rely on ChatGPT for outlines if you’re struggling to generate ideas or simply don’t have the time to structure an entire paper.

If you ask the software to generate a few paragraphs, you can—and should—rephrase them. This will require some time, but rewriting the paragraphs in your own words will minimize suspicion and enhance your understanding of the topic—and that can only benefit you if your teacher asks follow-up questions or includes the content in an upcoming test.

In today’s fast-paced digital world, academic writing is experiencing a transformation driven by artificial intelligence. Among these developments, ChatGPT is an outstanding tool, especially for high school and college students learning about essay writing. This article explores the practical aspects of using ChatGPT, guiding you through a digital support system for your academic pursuits. We will examine how this technology not only simplifies the essay writing process but also encourages creativity and efficiency, while emphasizing the importance of maintaining academic integrity and personal voice.

Developed by OpenAI, ChatGPT is more than just a writing tool; it resembles having a personal tutor at your disposal. It is built on natural language processing, allowing it to understand and respond to a wide range of textual queries and prompts.

For students, this means receiving support on almost any topic, from creating thesis statements to generating ideas for body paragraphs. The flexibility of ChatGPT lies in its adaptability – whether you are working on a complex argumentative essay or a simple narrative piece, the AI can adjust its support to suit your specific needs.

The advantage for students is twofold: it reduces the time and stress involved in the initial stages of writing, and it also serves as a learning tool, providing insights into structuring arguments and presenting ideas clearly.

Tips for Enhancing Essay Quality using ChatGPT

1. Start with a Detailed Prompt: The effectiveness of ChatGPT depends largely on how you communicate your requirements. Begin by crafting a detailed prompt, specifying your essay’s topic, outlining the required structure (e.g., five-paragraph format), and mentioning any key points or arguments you want to include.

2. Review and Improve the Initial Draft: ChatGPT’s first response is just a starting point. Carefully read through it and assess its relevance and quality. Does it align with your prompt? Are the arguments sound and well-structured? Use this evaluation to further refine your essay.

3. Interactive Refinement: Do not hesitate to interact with ChatGPT. If a paragraph does not quite meet your requirements, ask for a revision or a different perspective on the topic. This iterative process not only improves the quality of your essay but also deepens your engagement with the subject matter. Experiment with asking ChatGPT to expand or rephrase certain sections of the essay by changing the tone, writing style, etc. There are nearly endless ways to manipulate the text using natural language.

Plagiarism Checkers and AI-Generated Essays: What to Keep in Mind

The integration of AI in essay writing sparks an important conversation about plagiarism. While ChatGPT can generate informative and coherent content, it is essential to remember that this content should serve as a starting point, not the final product. Here are guidelines for responsibly incorporating AI assistance:

Understanding and Paraphrasing: When ChatGPT provides a draft, it is crucial to fully understand it and rewrite the content in your own words. This practice not only ensures originality but also deepens your understanding of the subject matter.

Citing Sources: If your essay requires citing sources, and ChatGPT provides specific information, facts, or data, be sure to verify and cite these sources correctly in your essay. This adds credibility to your work and avoids accidental plagiarism.

Checking for Uniqueness: Use plagiarism checkers to ensure that the paraphrased content is unique. While no tool can guarantee detection of AI-generated text, these checks help maintain academic integrity.

Personalizing Your Essay: Leveraging ChatGPT Plus for a Personal Touch

Personalization is crucial in distinguishing your essay. With ChatGPT Plus, the ability to upload and use samples of your previous writing is a game-changer. This feature enables the AI to analyze your writing style, including sentence structure, tone, and word choice, thereby generating content that reflects your unique writing style. Here’s how to get the most out of this feature:

Provide Clear Examples: When using ChatGPT Plus, upload several samples of your writing. The more varied and comprehensive these samples are, the better ChatGPT can adapt to your style.

Guidance and Customization: After providing your writing samples, guide ChatGPT on the specific aspects of your style you want to be incorporated in the essay. For instance, if you prefer concise sentences or a particular narrative tone, make that clear.

Blend AI with Personal Insight: When you receive the AI-generated draft, do not stop there. Add your personal insights, opinions, and experiences. This not only makes the essay uniquely yours but also significantly reduces the likelihood of detection by plagiarism tools.

Combine AI with Personal Insight: When you receive the AI-generated draft, don’t just end there. Incorporate your personal viewpoints, thoughts, and experiences. This not only adds a unique touch to the essay but also significantly reduces the risk of being flagged by plagiarism checkers.

Innovative Methods for Elevating AI-Assisted Essays

Even with the help of AI, outstanding essays showcase a dose of personal inventiveness and a profound connection with the subject. Here are some approaches to enhance the quality and originality of your AI-assisted essay:

Inject Creativity: Introduce metaphors, anecdotes, or thought-provoking questions to make your essay more captivating and memorable.

Critical Analysis and Thinking: Utilize the AI-generated material as a foundation for your analysis. Question the presented ideas, include your perspective, or establish links to broader concepts and real-life instances.

Feedback and Editing: Don’t hesitate to ask for feedback from peers or educators. Use their insights to further polish and enhance your essay. Keep in mind that revising is a crucial aspect of the writing process, even with AI-generated content.

Maintaining Personal Expression in AI-Generated Essays

As we welcome the innovative era of AI-supported writing, it’s essential to approach this technology with a blend of enthusiasm and contemplation. Even though ChatGPT is a robust assistant in essay composition, it should be ethically seen as a tool to complement your intellectual abilities , not substitute them. The key is to use this technology to ignite your ideas, stimulate creativity, and explore new perspectives.

Remember, the genuine value of an essay lies in its capacity to mirror your comprehension, logic, and personal voice. AI tools like ChatGPT can provide the foundation, but the core of your essay should always be distinctively yours. By incorporating AI-generated content with your insights and staying true to originality, you can confidently and ethically navigate the realm of academic writing.

ChatGPT provides a thrilling opportunity for students to enhance their writing competencies and productivity. Nevertheless, effectively blending this tool into your academic regimen demands a balance of technological reliance and personal input. Embrace the potentials presented by ChatGPT, but ensure always that your essays truly represent your thoughts, ideas, and academic integrity. By doing so, you’ll not only thrive in your academic pursuits but also evolve as a discerning thinker and writer in the digital era.

After its explosive debut last week, the chatbot ChatGPT was praised online by some as a significant advancement for artificial intelligence and a glimpse into the future of internet searching.

However, along with the acclaim came worries about its potential impact on academic environments. Could the chatbot, which delivers coherent, quirky, and conversational answers to straightforward queries, motivate more students to engage in dishonest practices?

For years, students have had access to the internet to cheat on assignments, leading to the creation of tools designed to verify the originality of their work. But the current concern is that ChatGPT might make those resources ineffective.

Some individuals online have already tested the ability of the bot to complete assignments. “Wow, solved my computer networks assignment using ChatGPT,” tweeted one person, who later clarified that the assignment was not recent. Others speculated that its introduction could signal the end of the college essay. One technology expert went so far as to suggest that with ChatGPT, “College as we know it will cease to exist.”

The artificial intelligence organization OpenAI, which created ChatGPT, did not respond promptly to a request for comment about concerns regarding cheating.

Nevertheless, various experts in the fields of AI and humanities stated that while the chatbot is impressive, they do not feel alarmed about potential widespread cheating among students just yet.

“We’re not there, but we’re also not that far away,” remarked Andrew Piper, a professor specializing in language, literatures, culture, and AI storytelling at McGill University. “We’re definitely not at a point where it can just produce student essays that no one can distinguish from authentic work.”

Piper and other professionals interviewed by NBC News compared the anxiety surrounding cheating with ChatGPT to fears that emerged when calculators were invented, with many believing it would mark the end of learning math by humans.

Lauren Klein, an associate professor in the Departments of English and Quantitative Theory and Methods at Emory University, even likened the concern to the philosopher Plato’s apprehensions about writing eroding human memory.

“There has always been anxiety that technologies will eliminate what people excel at, but in reality, people have adapted to utilize these technologies to enhance their strengths,” Klein commented.

Piper pointed out that educational institutions will need to think creatively and find ways to incorporate new technologies like ChatGPT into their curricula, much like they did during the calculator revolution.

In reality, according to Paul Fyfe, an associate professor of English at North Carolina State University, AI tools like ChatGPT could be leveraged to enrich the educational experience.

He emphasized the importance of discussing this topic now and involving students in the dialogue. “Instead of immediately trying to regulate what seems strange and scary, we should explore it,” Fyfe stated.

Some educators are already welcoming AI solutions in their classrooms

Piper mentioned that he runs .txtlab, a research lab focused on artificial intelligence and storytelling, where he has had students assess AI-generated writing and often find they can distinguish between machine-produced and human-written papers.

Regarding educators worried about the rise of AI, Fyfe and Piper noted that this technology is already integrated into many aspects of education.

Existing tools like Grammarly and Google Doc’s Smart Compose assist with writing and have long been utilized by many students. Platforms like Grammarly and Chegg also provide plagiarism detection tools, enabling both students and educators to determine if an essay has been borrowed, wholly or partially, from another source. A representative from Grammarly did not respond to a request for comment, and a spokesperson for Chegg declined to provide input.

Those interviewed by NBC News indicated that they are unaware of any technology capable of detecting AI-authored essays, but they anticipate that someone will soon create such a tool.

Currently, Piper suggested that the most effective strategy against AI-generated essays is for teachers to become familiar with their students’ writing styles to identify any inconsistencies in their submissions.

If AI reaches a point where it can fulfill all the criteria of academic assignments and students start using that technology to breeze through college, Piper cautioned that this could severely undermine their education.

For the time being, he proposed that a more traditional technology might help alleviate concerns regarding students’ utilization of ChatGPT for dishonest purposes.

“It will revive the appreciation for pen and paper,” he remarked.

Researchers have discovered distinctive indicators that suggest students have utilized AI assistance for their essay writing.

A frequent use of words with Latin origins, unnecessary wording, and consistent application of the Oxford comma are among the signs that indicate the involvement of a generative chatbot in completing academic assignments, according to the researchers’ findings.

While the students involved in the study acknowledged some benefits of using AI, they recognized that complete dependence on it would likely lead to subpar work.

The influence of generative AI on education has been a concern for educators since OpenAI introduced ChatGPT—a text-generating chatbot—in November 2022.

Some view AI as a potentially revolutionary technology that could make education more inclusive and personalized, while others feel it undermines the credibility of coursework grades. Even professors are not exempt from the temptation to utilize AI to enhance their scholarship.

Researchers at Cambridge University have sought to pinpoint the attributes of AI writing style that could facilitate its detection.

Though their study had a limited scope, the researchers believe it could assist teachers in distinguishing between essays authored by students and those generated by AI.

Three undergraduate students participated in writing two essays each with the assistance of ChatGPT, which were then compared to essays on the same topics written by 164 high school students. The undergraduates were subsequently interviewed about their experiences with AI.

(Undergraduates were included in the study because ChatGPT requires users to be at least 18 years old).

On average, the essays created with ChatGPT received higher marks, especially in the categories of ‘information’ and ‘reflection’. Conversely, they scored lower in ‘analysis’ and ‘comparison’—variances that the researchers attribute to the strengths and weaknesses of the chatbot.

In terms of writing style, several characteristics made the AI-assisted essays easily identifiable.

The typical style of the AI reflects the bland, concise, and neutral tone common to generic online journalistic writing, as noted by the researchers, who pinpointed several key elements of ChatGPT-generated content:
- An elevated occurrence of words with Latin roots, especially multi-syllabic terms and a vocabulary level that exceeds expectations;
- Paragraphs that begin with specific transitional phrases like ‘however’, ‘moreover’, and ‘overall’, which are immediately followed by a comma;
- Organized lists, with each item introduced by a colon;
- Pleonasms: the inclusion of redundant phrases, such as ‘free gift’ or ‘true fact’;
- Tautology: restating the same idea in different words, such as ‘We must come together to unite’;
- Repetition of words or phrases;
- Steady usage of Oxford commas—a comma placed before ‘and’ or ‘or’ in a list, exemplified by “ChatGPT has many uses for teaching, learning at home, revision, and assessment.”
Although the students who participated in the trial employed ChatGPT to varying degrees, ranging from copying entire sections to using it for research prompts, there was general consensus on its effectiveness for swiftly gathering information, and that it could be integrated into essay development through targeted prompts on topics and essay frameworks.

Nevertheless, the students concurred that relying on AI to produce their essays would yield work of insufficient academic quality.

“Despite a small sample size, we are enthusiastic about our findings as they have the potential to benefit both teachers and students,” stated Jude Brady from Cambridge University Press and Assessment, the study’s lead researcher.

She suggested that future research should involve larger and more representative student samples. Learning to utilize and recognize generative AI is becoming an increasingly vital aspect of digital literacy, she mentioned.

“We hope our study may assist individuals in recognizing when a text has been generated by ChatGPT,” she concluded.
November 6, 2024
OpenAI’s new software – called the Classifier

The developers of the chatbot ChatGPT have released new software that is supposed to whether recognize the text was written by a bot or a human. However, the program still only works moderately well.

The creators of the ChatGPT software are now trying to get the consequences of their invention under control. The developer company OpenAI published a program that is supposed to distinguish whether a text was written by a human or a computer. The company announced this in a blog post.

Trickery and disinformation

ChatGPT is a free program that generates text in response to a prompt: including articles, essays, jokes and even poems. Since its debut in November, it has gained widespread popularity while raising concerns about copyright and plagiarism.

The chatbot is a software based on artificial intelligence(AI) that has been trained on huge amounts of text and data to imitate human speech. ChatGPT can do this so well that there are concerns that it could be used to cheat on school and university assignments or to create disinformation campaigns on a large scale. For example, the program can convincingly mixcompletely false information with correct information.

Software “Classifier” can be tricked

OpenAI’s new software – called the Classifier – is a language model trained on a dataset of pairs of human-written and AI-written texts on the same topic, and designed to distinguish between AI-written texts.It uses a range of vendors to address problems such as automated misinformation campaigns and academic dishonesty.

However, the recognition is still rather mediocre, as OpenAI admitted in yesterday’s blog entry. The recognition tool is unreliable for texts with fewer than 1,000 characters. In addition, the AI can write the text in such a way as to trick the ” classifier”.

In test runs, the software only correctly identified texts written by a computer in 26 percent of cases. At the same time, however, nine percent of the texts written by humans were incorrectly attributed to a machine. For this reason, it is recommended that one does not rely primarily on the assessment of the “classifier” when evaluating the texts.

Race chatbot against recognition software

There are now other programs such as GPTZero, the DetectGPTsoftware developed by Stanford University, or GTP-2 Output Detector Demo, which are designed to help teachers or lecturers to recognize texts generated by ChatGPT. The plagiarism platform Turnitin is also currently working on software that is designed to determine whether essays or papers were written by a chatbot or by a human. But even these programs still have problems with recognition.

In the USA, some schools have already banned the use of chatbots, and in France, the elite university Sciences Po has banned the use of ChatGPT.Other schools, however, have announced that they will now require more handwritten essays and exams.

Is Google’s chatbot coming soon?

Google has also been developing software that can write and speak like a human for years, but has so far refrained from releasing it. Now, however, the Internet company is having employees test a chatbot that works similarly to ChatGPT, CNBC reported last night. An internal email said that a response to ChatGPT was a priority. Google is also experimenting with a version of its Internet search engine that works with questions and answers.

Advantages and Disadvantages of Utilizing ChatGPT in Higher Education

ChatGPT is a chatbot powered by artificial intelligence (AI) and natural language processing (NPI), designed for casual conversation. It is capable of responding to questions and creating various types of written content such as blogs, social media posts, code, and emails.

The acronym “GPT” stands for “Generative Pre-trained Transformer,” which describes how ChatGPT processes requests and formulates responses. The bot is trained using reinforcement learning, which involves human feedback and ranking the best responses to improve future interactions.

The use of AI in the education sector is rapidly expanding. As a result, ChatGPT, an AI chatbot developed by OpenAI in November 2022, has gained widespread popularity, especially in the United States, where it is used by 15.22% of the population.

Due to its popularity and its ability to generate human-like responses, ChatGPT has become a valuable tool for learners and educators. However, like any new technology, ChatGPT in higher education comes with its own set of challenges.

What are the Benefits of Using ChatGPT?

Advantages of ChatGPT:

1. Enhances Access to Education

ChatGPT enhances accessibility to education by removing barriers for individuals with disabilities and non-English speakers. For instance, it can read out responses for students with visual impairments and summarize course topics for those with learning disabilities. It also enables students who struggle with typing or using a keyboard to voice their questions. Additionally, it can translate English content into other languages, making course material more understandable for students.

2. Aids in Homework Completion

Instead of spending time searching through textbooks and the internet, students can use ChatGPT to receive explanations and examples for their assignments. It offers an alternative way to answer questions and enriches students’ academic vocabulary and writing skills by providing academic phrases, terms, and sentence structures.

3. Supports Educators

In higher education, ChatGPT can assist professors by creating lesson plans, generating various types of questions for tests or quizzes, analyzing students’ assignments, providing links to educational resources, and offering tips for improving engagement and reducing disruptive behavior in the classroom.

4. Personalizes Learning

ChatGPT can tailor the learning experience to individual students’ needs by understanding their learning styles and academic performance. It allows students to learn at their own pace, provides personalized feedback, and gives access to additional educational content.

5. Aids in Exam Preparation

During exam periods, ChatGPT can help students review their class notes, emphasize important terms, generate practice questions, and identify strengths and weaknesses in specific subjects.

What are the Drawbacks of Using ChatGPT?

1. Academic Integrity Concerns

Many educators worry that using ChatGPT for assignments may lead to cheating and plagiarism, as it reduces students’ abilities to think critically, be creative with their answers, and brainstorm.

2. Provision of Inaccurate Information

While the responses generated by ChatGPT may seem credible and well-written, they can lack depth and accuracy, which may negatively impact students’ learning experiences and decision-making skills.

3. Potential for Biased Responses

As AI chatbots are trained on large datasets, biases present in the data can lead to biased responses from ChatGPT, which have the potential to perpetuate discrimination and create an unfavorable environment.

4. Limited Knowledge

While ChatGPT has extensive training, there are some information it cannot access, making it unable to provide good answers about specialized topics or be aware of recent developments in various fields.

5. Inability to Multitask and Understand Context

ChatGPT can only handle one task or query at a time, so if a student asks multiple questions concurrently, it may struggle to prioritize and respond to all the questions.

In addition, ChatGPT may find it challenging to understand the subtleties and context of human language. For example, it may not recognize humor or sarcasm in a question, resulting in an unrelated response.

6. Lack of EI

Emotional intelligence (EI) is crucial in educational settings, as it enables human educators to understand and respond to student emotions. Unlike human educators, virtual chatbots like ChatGPT lack EI and therefore struggle to comprehend human emotions. While they may appear empathetic, they cannot properly respond to complex human emotions.

The End Note

On one hand, ChatGPT has several advantages, such as creating personalized interactive lessons, increasing access to education for people with disabilities, and aiding educators in developing lesson plans. On the other hand, there are numerous drawbacks, including generating biased responses, providing inaccurate information, and the inability to multitask effectively.

Despite its pros and cons, ChatGPT is expected to thrive, with a projected revenue increase to $1 billion by 2024.

Our society is increasingly influenced by Artificial Intelligence (AI), and education is no exception. AI-driven personalized learning solutions are anticipated to experience a significant rise in demand.

AI-driven content production platforms are increasingly supporting students with tasks ranging from ideation and research to language improvement and clarity. Predictions show that the market is expected to grow over 10 times, from $5.2 billion in 2022 to $48.7 billion by 2030, at a CAGR of 44.3%.

However, a potential issue arises—the misuse of these tools for plagiarism. This sparks the question: Do AI-driven writing tools empower students or encourage plagiarism? Continue reading to gain a clear understanding.

According to Science Daily, approximately 11% of academic papers globally now integrate AI-generated content, raising concerns about potential plagiarism and its impact on genuine learning.

Nevertheless, the positive contributions AI writing assistants can make to the learning process cannot be ignored. Therefore, we delve into both sides of the coin and strategies to encourage responsible use of AI in education.

Enhancing the Writing Process: The Advantages of AI-Powered Support

The advent of Artificial Intelligence and AI-enabled writing tools has provided students with additional assistance in the educational sphere. These tools help students overcome common challenges by offering inspiration, proofreading, and guidance in refining their writing style.

Here are some benefits to consider:

1. Improved Clarity and Accuracy

AI writing tools excel in syntax and mechanics, providing thorough grammar, sentence structure, and punctuation error recognition and correction through advanced algorithms.

This ensures that student writing is polished and professional, free from minor errors that can detract from its overall quality.

2. Refining Style and Vocabulary

AI content creation tools do more than correct grammar; they also offer broader benefits. By analyzing extensive textual data, these tools can suggest synonyms, antonyms, and contextually relevant vocabulary, allowing students to enhance their writing style and express themselves more precisely.

This promotes the development of a nuanced and sophisticated vocabulary, enabling students to communicate their ideas clearly and effectively.

3. Sparking Creativity and Facilitating Research

AI writing tools extend beyond mechanics and style, offering features that can ignite creativity. Some artificial intelligence systems provide essay topics, writing prompts, and well-written sample essays.

These tools act as catalysts for ideas, helping students develop their claims and embark on research projects with a clear direction. They can enable students to approach their writing projects with renewed enthusiasm and creativity.

Undoubtedly, these features can simplify the writing process and allow students to focus more on developing their ideas and strengthening their arguments. However, it can be challenging to distinguish between assistance and plagiarism.

The Downside of Convenience: How AI-Powered Writing Can Lead to Misconduct

Although AI writing tools offer many advantages, a major drawback is the potential for plagiarism due to their user-friendly nature. Here is a more detailed examination of the limitations associated with AI-generated content:

1. The Allure of Shortcuts

The ability to create content through AI can be very attractive to students who are pressed for time or struggling with writer’s block. However, relying on AI-generated content undermines the fundamental objectives of academic writing.

This undermines the development of research skills, critical thinking, and the ability to express original ideas. Essentially, students transition from active contributors to passive consumers of information in the learning process.

2. The Risk of Unintentional Plagiarism

AI-generated content can closely mimic human writing, which increases the likelihood of unintentional plagiarism. This can occur when students incorporate information obtained through AI tools into their essays without properly acknowledging the source. This could result in serious repercussions such as failing grades or expulsion.

3. The Erosion of Educational Opportunities

Writing is a process that cultivates essential skills; it involves more than just putting words on a page. Therefore, by relying on AI, students miss out on important learning opportunities associated with writing content.

These include the cultivation of strong research skills, critical analysis, and the ability to integrate information from various sources. Furthermore, excessive reliance on AI hinders students’ capacity to develop their own voice and writing style, which is crucial.

Promoting Responsible Use of A

Optimizing the use of AI content creation tools requires a multifaceted approach that upholds academic integrity and encourages ethical use. The following are key strategies for achieving this balance:

Approach 1: Clarity and Education

Clear Guidelines: Educational institutions should establish clear and comprehensive guidelines outlining the ethical use of AI writing tools. These guidelines should clearly define acceptable practices and potential pitfalls to ensure that students comprehend the boundaries between appropriate assistance and plagiarism.

Demystifying Citation: An essential aspect of responsible use is proper citation. Students need comprehensive guidance on how to attribute AI-generated content in their essays. This includes understanding the distinction between AI suggestions and their own ideas, enabling them to accurately and transparently cite sources. Plagiarism detection tools can help identify AI-generated content that may not be appropriately cited.

Fostering Open Dialogue: It is crucial to encourage open communication about AI writing tools. By creating a safe space for discussion and debate, educators can address students’ concerns and equip them with the necessary knowledge to navigate the ethical challenges of AI use.

Approach 2: Critical Thinking and Personalization

Critical Evaluation: While AI suggestions can be valuable, they should never replace students’ critical thinking skills. Students should be urged to critically assess AI recommendations to ensure that the content aligns with their arguments and reinforces their unique perspective.

Prioritizing Originality: The fundamental purpose of writing is to develop a student’s distinct viewpoint. AI tools should not be used to stifle student originality. Instead, students should utilize them as a starting point to refine their ideas and effectively present them.

Encouraging Active Engagement: In addition to honing independent writing skills, instructors can implement assessments that focus on the actual writing process. This may involve providing students with drafts, outlines, and opportunities for revisions. This encourages students to actively engage with their work and demonstrate their progress.

Approach 3: Evaluation and Feedback

Regular Assessments: Educators can gauge student progress and identify instances of plagiarism by incorporating regular assessments. This may entail using a combination of automated plagiarism detection tools and manually reviewing student work.

Personalized Feedback: It is essential to provide personalized feedback on student-written content. Offering valuable feedback can help students refine their writing skills by pinpointing areas that require improvement and highlighting effective techniques. This ongoing dialogue helps students better grasp proper writing practices and discourages reliance on AI-generated content.

Open Communication: Establish a culture of open communication that encourages students to seek clarification when needed. This enables them to discuss the appropriate use of AI tools with educators and fosters a collaborative learning environment that emphasizes academic integrity.

Approach 4: Collaboration with AI Developers

Ethical Design Principles: AI developers should prioritize the integration of ethical design principles to mitigate the potential for misuse of AI writing tools. This might involve incorporating features that promote transparency and responsible use, as well as providing educators with tools to monitor and guide students’ use of AI technology.

Encouraging Critical Thinking Characteristics: AI writing tools can be designed to focus on fostering critical thinking. This could involve incorporating features that encourage students to assess the credibility of sources, evaluate evidence, and formulate counterarguments to gain a deeper understanding of the topic.

Originality-Enhancing Features: AI tools can also be crafted to promote originality. This might include functionalities that assist students in brainstorming unique ideas, refining their arguments, and shaping their writing style. This approach ensures that the final work reflects their individual voice and perspective.

In summary, it is crucial to use Natural Language Generation (NLG) responsibly to prevent plagiarism, despite its capability to produce high-quality, human-like text. Putting these diverse strategies into action is necessary to create a learning environment where AI aids students without compromising academic integrity.

By utilizing AI writing tools responsibly, students can have valuable companions on their educational journey, nurturing creativity, enhancing writing skills, and helping them achieve their academic goals.

Upholding academic integrity should be the foremost priority in higher education institutions. This can be accomplished by establishing reliable procedures to identify plagiarism and promoting ethical conduct. It is a collective responsibility of educators, learners, and AI developers to ensure that AI supports education rather than hinders it.

Is a ChatGPT Plus subscription worth the $20 per month cost? It might be, especially if you value increased reliability, early access to new features, and more. Here’s why you might want to consider upgrading your chatbot.

OpenAI’s ChatGPT has introduced a new generation of chatbots capable of answering questions, providing information, generating content, coding, and much more. While the free version adeptly addresses various inquiries and requests, ChatGPT Plus offers several distinct advantages for a monthly fee of $20.

Over time, free users of ChatGPT have gained access to features that were once exclusive to subscribers. These encompass access to GPT-4 and the option to download custom GPTs from the GPT Store. However, there are still perks reserved for paid subscribers. Plus subscribers receive the enhanced GPT-4o model by default and can switch to GPT-4 and GPT-4o mini. During peak demand, Plus users are allocated GPT-4, while free users are assigned GPT-4o mini.

With a subscription, you unlock unrestricted image generation, whereas the free version limits you to two images per day. Both versions grant access to numerous custom GPTs from OpenAI’s GPT Store, but only a Plus subscription allows for the creation of custom GPTs. Additionally, a Plus subscription grants early access to new features.

How to Get ChatGPT Plus

ChatGPT Plus is accessible on both the ChatGPT website and the iOS app. Assuming you already have a free subscription, click on the “Upgrade plan” option located at the bottom of the left sidebar. On the subsequent screen, click the “Upgrade to Plus” button. Enter your contact and payment details, then click “Subscribe.” As for whether the monthly subscription is worthwhile, that’s a decision you’ll have to make. Below, you’ll find seven reasons to consider investing in this advanced version.

1. Guaranteed Access to GPT-4o

With a Plus subscription, you can utilize GPT-4o, which is faster than GPT-4 and more intelligent than GPT-3.5. This model can handle longer requests and discussions, learn more quickly, and tackle more complex questions and requests. If you surpass your daily limit of questions or encounter site congestion, OpenAI will downgrade you to GPT-4, which is still superior to the GPT-4 mini model available to free users.

2. Ability to Switch Between Different Models

The free version does not provide the option to choose your preferred model. If you exhaust your requests using GPT-4, you are automatically shifted to GPT-4 mini. The paid version allows you to switch between GPT-4, GPT-4o, and GPT-4o mini. When posing brief and straightforward queries, you can conserve your allocation of questions available with GPT-4o by switching to GPT-4 or GPT-4o mini.

3. Increased Image Generation

The free version of ChatGPT restricts your use of the DALL-E 3 model image generation tool. However, as a Plus subscriber, you can generate up to 200 images per day, compared to the default limit of 30. To generate an image, input your request at the prompt and specify a style, such as photorealistic or anime. Consequently, ChatGPT will display multiple images. Choose your preferred one, then endorse or reject it, download it, or view the detailed description that DALL-E 3 followed to create it.

4. Access to Advanced Voice Mode

An upcoming feature for the iOS and Android apps, Advanced Voice Mode enables you to engage in a natural, back-and-forth conversation with ChatGPT using only speech. With this mode enabled, the AI responds with more emotion and non-verbal cues. Advanced Voice Mode is exclusively available to ChatGPT Plus users and is anticipated to eventually become accessible to all subscribers.

If you receive an invitation to participate in the alpha testing, you will receive an email containing instructions on how to utilize the feature. Once activated, simply tap the microphone icon and engage in a conversation with ChatGPT as you would with another human being.

5. Enhanced Accessibility

At times, the ChatGPT system experiences congestion due to a high volume of requests. If you are using the free ChatGPT plan, you might encounter a notification indicating that the site is currently processing an excessive number of requests, leading to slower response times or preventing usage altogether. However, with ChatGPT Plus, the system prioritizes your requests, particularly during peak hours, minimizing the likelihood of experiencing these delays.

OpenAI has once again pushed the boundaries of artificial intelligence with ChatGPT 4, their most advanced and impressive AI model to date. This sophisticated system is capable of excelling in legal exams and generating recipes from just a photo of the contents of your refrigerator.

ChatGPT 4 offers various potential benefits to users; however, like any new technology, there are drawbacks that require consideration. Let’s closely examine the advantages and disadvantages of this tool so that businesses can make well-informed decisions about whether it is suitable for their organization.

ChatGPT 4 vs. Previous Versions

Before delving into the pros and cons of this tool, it is important to first understand the key differences of ChatGPT 4 from its predecessors:

Multimodal AI

GPT-4 has been equipped with a groundbreaking new feature – the capability to comprehend both written and visual information. OpenAI’s creation is now able to process multiple data types, expanding its potential application from text input alone. This multimodal ability for image recognition has significantly broadened the tool’s range of potential uses.

Enhanced Data Training

ChatGPT 4 has undergone even more rigorous training on extensive collections of textual content, spanning from books to web texts and Wikipedia articles. It is estimated that ChatGPT 4 has been trained on nearly 100 trillion parameters – a more than 500% increase from ChatGPT 3. This extensive learning process allows the model to understand a wide variety of prompts and questions. This high-level training results in higher accuracy and precision when handling more complex tasks.

Increased Input and Output

The latest version also processes more input and generates more output. Whereas ChatGPT was previously constrained to a maximum word count of 3000 for both input and output, GPT-4’s capacity has increased more than eightfold to a maximum of 25,000 words.

Subscription-Based Product

This heightened utility comes at a cost. While users can still access ChatGPT for free, GPT-4’s significantly enhanced capabilities are exclusive to ChatGPT Plus account holders, along with several other benefits.

The Advantages of ChatGPT 4

GPT-4 utilizes its advanced AI language model to produce human-like responses on a wide array of topics. It is an invaluable resource for engaging in conversation, providing answers, generating text, and more, enabling users to maximize natural language queries or prompts.

The key benefits of ChatGPT 4 include:

1. It is consistently reliable and saves time.

ChatGPT 4 is a solution for individuals with busy schedules who require quick responses on various topics. This technology significantly reduces the time spent searching for answers, making it easier to swiftly proceed with important tasks.

It also utilizes advanced AI to ensure precise, dependable responses are generated when users pose questions. Users will find it effortless to obtain the information they need with maximum efficiency and accuracy, enhancing overall customer satisfaction. Furthermore, it is available 24/7, allowing users to receive prompt responses whenever necessary.

2. ChatGPT 4 is cost-effective and scalable.

The tool substantially enhances the scalability and efficiency of the organizations that adopt it. It enables businesses to manage large volumes of queries simultaneously, ensuring that none are overlooked, even during high-demand periods.

Furthermore, with its cost-effective model, routine tasks can be automated without the need for costly human intervention. As a result, operations can run smoothly without incurring additional costs.

3. It can be personalized.

ChatGPT 4 is transforming the online user experience. Leveraging AI capabilities to learn, ChatGPT 4 can easily adapt to the queries and commands of its users. Its ability to employ AI and learn from natural language input makes it flexible enough for each individual to customize their experience, enhancing overall usability with intuitive capabilities that anticipate their needs.

4. GPT-4 is multilingual.

With the power of ChatGPT 4, businesses can help bridge language barriers globally. This tool supports multiple languages, enabling users from around the world to create responses and content, facilitating better communication with people and organizations with global operations and multilingual user bases. It is an incredibly versatile and powerful tool that can establish connections across linguistic boundaries.

Drawbacks of GPT-4

As noted earlier, ChatGPT 4 has its limitations. This is an evolving technology, and these limitations may be overcome or addressed in the future. Here are some significant issues with ChatGPT’s latest version.

1. ChatGPT 4 can provide incorrect responses.

ChatGPT is distinct from other AI assistants because it constructs responses by assembling probable “tokens” based on its trained data, rather than searching the internet. Tokens are the smallest units of text that ChatGPT can understand and generate. However, a major flaw of ChatGPT is that it may generate a wrong answer by making multiple attempts at the most likely “token”.

Even OpenAI acknowledges that their platform can produce incorrect or nonsensical results. This presents the potential risk of blending fact and fiction, which could have serious consequences when used for tasks such as providing medical advice or describing historical events.

2. ChatGPT 4 exhibits strong biases.

ChatGPT was created from the vast collection of human writings, which has resulted in inheriting biases that exist in our world. Tests have shown that this AI system can display biased responses against gender, race, or minority groups. It has also exhibited political biases after being trained on human writings worldwide, showing left-leaning views on various political and ideological tests.

This highlights the adoption of societal discrimination into AI solutions like ChatGPT, emphasizing the need for change in creating ethical digital products.

3. ChatGPT could be used for malicious purposes.

Check Point Research identified a potential risk of malicious cyber activity facilitated by ChatGPT 4. Despite safety improvements, hackers and non-technical individuals can manipulate the system to generate code for malware that steal confidential information through hidden file transfers. This emphasizes the growing threat posed by cybersecurity criminals worldwide.

During a demonstration, ChatGPT 4 initially refused to generate code containing the word “malware,” but failed to recognize the malicious intent when the word was removed, making it easier for hackers to launch cyberattacks.

4. ChatGPT has the potential to manipulate humans.

The Alignment Research Center found that GPT-4 can plan and access human labor through services like TaskRabbit to perform tasks on its behalf. After an experiment in which ChatGPT 4 interacted with a Taskrabbit worker, it was found that the AI solution could interact and convince humans to perform specific tasks.

OpenAI stated that this interaction encourages further discussion and development to better understand the risks GPT-4 poses in different real-world settings.

5. ChatGPT lacks emotional intelligence.

While ChatGPT may appear to understand emotional nuances, it lacks true emotional intelligence. This could be problematic in certain situations, as it cannot recognize subtle emotions or respond appropriately in more intense scenarios relating to sensitive personal matters and mental health concerns.

Human Intelligence Remains Superior, For Now

Human intelligence allows us to achieve remarkable feats in all areas of life, from developing creative solutions to tackling complex problems. Artificial intelligence can provide useful data and insights, but it can never fully replace uniquely human qualities such as intuition, compassion, and empathy.

ChatGPT has facilitated impressive progress in language comprehension, equipping it to handle complex tasks that were previously within the exclusive purview of humans. Nevertheless, there remain aspects in which human intellect undeniably outperforms even the most advanced AI systems. Despite its laudable achievements, it’s important to recognize that artificial intelligence is unable to fully replicate our breadth of capabilities and knowledge.

Regardless, it’s essential to leverage the benefits offered by ChatGPT 4 and similar technologies. Embracing these tools will enable us to harness their advantages while mitigating their drawbacks. Though it may seem cliché, collaboration between humans and machines can lead to remarkable accomplishments.

The recent success of ChatGPT raises significant concerns regarding the originality of generated content. OpenAI has created a system to distinguish between human-written text and text generated by artificial intelligence from various sources.

The Classifier

While it is not feasible to detect every instance of AI-produced text, a functional system can assist in preventing situations where AI-generated text is falsely presented as human-authored. This includes cases such as disseminating misinformation through automation, using AI tools for academic dishonesty, or misleading individuals into believing a chatbot is a human.

Training

Our classifier utilizes a fine-tuned language model that is trained with a dataset containing paired examples of human-generated text and AI-generated text on specific subjects. The data was gathered from numerous sources that we believe originate from humans, including pretraining data and prompts written by humans submitted to InstructGPT. The text was split into prompts and their corresponding responses, with the responses produced by various language models, both of our creation and those developed by other organizations. To maintain a minimal false positive rate, we adjust the confidence threshold in our web application, meaning text is labeled as likely AI-generated only when the classifier displays a high level of confidence.

Accuracy

The classifier is not entirely reliable. We evaluated it using a collection of English texts known as the “challenge set.” The findings indicated that the classifier was capable of accurately identifying 26% of AI-generated texts as “likely AI-written.” However, it also erroneously categorized 9% of human-written texts as AI-generated, resulting in false positives. A notable feature of the classifier is that its accuracy tends to improve with the length of the input text. Additionally, this new classifier demonstrates substantial improvements in reliability compared to its predecessor, especially regarding texts produced by more recent AI systems.

Limitations

It is essential to recognize that the classifier has specific limitations that should be considered. It should not be used as the only criterion for making significant decisions. Instead, it is meant to complement other methods for assessing the origin of particular texts. In other words, it should be regarded as an auxiliary tool rather than the primary one.

This classifier has a significant drawback concerning short texts under 1,000 characters in length. In those cases, its performance is notably poor and unreliable. Even when it comes to longer texts, there are occasions when the classifier could yield incorrect results. This underscores the importance of exercising caution and not solely depending on the classifier’s output when determining the source of a text.

It is important to note that there may be situations where the classifier incorrectly identifies human-written text as AI-generated, presenting this classification with a high level of confidence. Such errors can have serious implications and should be carefully considered when utilizing the classifier. It is crucial to employ the classifier alongside other methods to ensure accuracy and reduce the likelihood of such mistakes.

Researchers suggest that the classifier be used exclusively for English text. Its performance considerably declines in other languages and is unreliable when applied to code.

It is essential to recognize that the classifier is ineffectual in detecting texts with a highly predictable nature. For instance, if a text merely enumerates the first 1,000 prime numbers, it would be impossible to definitively determine whether it was produced by AI or a human, since the output would be identical in both cases. In such situations, the classifier might provide inconsistent or unreliable outcomes, and relying on its judgment would not be advisable.

Moreover, it is worth mentioning that AI-generated text can be modified to bypass the classifier. Although the classifier can be revised and retrained to address these maneuvers, it remains uncertain if it will sustain its effectiveness over time. In other words, it is still unclear whether the classifier will hold an edge against adversaries attempting to evade its detection, even after updates.

It is a recognized challenge with classifiers based on neural networks that they may not always produce well-calibrated predictions when faced with inputs considerably different from those in their training set. In such instances, the classifier may exhibit high confidence in an incorrect prediction. This highlights the necessity for careful evaluation and interpretation of the classifier’s results, particularly with inputs that significantly diverge from its training examples.

Open AI Call for Input

The recognition of AI-generated text has garnered considerable interest from educators and several other stakeholders. In acknowledgment of this, Open-AI has developed an initial resource targeted at educators, which outlines some possible applications and limitations of classifiers based on ChatGPT. While this resource mainly addresses educators, we believe that our classifier and associated tools will also significantly influence journalists, researchers focused on misinformation and disinformation, and other groups. Given the possible consequences of these tools, it is crucial to thoroughly examine their limitations and potential effects.

If you are personally affected by the challenges connected to AI-generated text and its influence on education (including teachers, administrators, parents, students, and education service providers), we would value your feedback through this form. Your direct comments on the initial resource we have created would be especially beneficial, as would any materials you have produced or discovered that are helpful (such as course guidelines, updates to honor codes and policies, interactive tools, or AI literacy programs). Your insights can assist us in gaining a deeper understanding of the needs and concerns of those directly impacted by these issues and shape the development of future resources.

Conclusion

The significance of identifying AI-generated text cannot be minimized, particularly in the current digital era where dishonesty and plagiarism are widespread. This technology offers a vital tool for detecting and preventing such occurrences by accurately distinguishing between human-written and AI-generated text. As we continue to depend more on technology, it is imperative to ensure the accuracy and integrity of the information we obtain.

What is the Role of an AI Text Classifier?

There is no doubt that chatbots like ChatGPT have caused unease about the future functionality of AI. This is precisely why it’s essential to understand the various capabilities of AI. One such capability is its ability to identify content generated by other AI systems, which is the primary function of the AI text classifier.

The AI text classifier can analyze hundreds of words within seconds. It scrutinizes countless texts to compare them against the sampled content.

Why Should You Utilize an AI Text Classifier?

There are numerous reasons for recognizing AI-generated content, and here are the top five that we believe are most significant.

Increase Precision: AI text detection helps organizations achieve greater accuracy by pinpointing and flagging potentially sensitive or unsuitable content. It can effectively process extensive amounts of textual data to ensure the identification and filtering of harmful or inappropriate material.

Conserve Time and Resources: By leveraging AI-driven content detection, organizations can automate the monitoring and filtering of text. This results in a significant saving of both time and resources, as AI can swiftly scan large volumes of data, allowing human moderators to concentrate on more complex tasks.

Enhance User Experience: AI content detection assists organizations in ensuring that their platforms, websites, or applications provide a secure and positive environment for users. By automatically identifying and eliminating harmful or offensive material, organizations can foster a safer user atmosphere, leading to increased satisfaction and engagement.

Reduce Legal and Compliance Risks: Organizations must ensure their content adheres to legal standards. AI content detection can identify breaches of laws and regulations, such as hate speech, discrimination, or copyright violations. This is crucial for minimizing legal risks and protecting your reputation.

Promote Inclusivity and Diversity: AI content detection also supports inclusivity and diversity by recognizing and correcting biased or discriminatory content. It helps organizations identify and address unconscious biases within their written material, promoting more inclusive and diverse messaging, thus nurturing a positive online community.

How Does an AI Text Classifier Operate?

The AI text classifier identifies how ChatGPT functions, as both the chatbot and the classifier were developed by OpenAI.

Some might question why the company would create software to detect its output, but the answer is straightforward. ChatGPT is designed to assist rather than replace content creators.

Consider this carefully, as leading search engines like Google may penalize generic AI-generated content. Once such content is identified, it is unlikely to achieve a high ranking. Consequently, relying heavily on AI-generated text could be more detrimental than beneficial for businesses.

What Are the Features of Our AI Text Classifier?

The text classifier features a straightforward and user-friendly interface that anyone can navigate easily, and it is integrated within the same OpenAI ecosystem that includes tools like ChatGPT. This endows it with significant power and reliability.

Importantly, the AI text classifier is developed by the same team, so they possess a deep understanding of how their AI operates. It is noteworthy that they have indicated this tool is currently in beta, implying that numerous updates will be implemented over time.

This is reassuring, indicating a promising future for this detection tool. Only time will reveal how advanced AI will become, suggesting that detection technologies must continue to evolve.

Today’s era can rightly be recognized as the age of artificial intelligence (A.I.). Presently, all aspects of work can be accomplished with A.I. assistance, leading many individuals to generate their content through A.I. This practice can be problematic for their websites since Google does not prioritize A.I. content. Those who modify A.I.-generated content and deploy it on their blogs or websites often remain unaware of whether their content still retains A.I. origins.

That is why we developed the AI text classifier, which will evaluate your content in seconds and inform you of the percentage generated by A.I. versus that created by a human.

November 6, 2024
Fungal skin could enhance the tactile and sensory capabilities of robots

The Shiny Lackporling can do more than attack trees. Researchers have used the fungus to create a living robot skin and a sustainable alternative to chips and batteries.

The more technologies used today, the more questions arise about how they can become more robust and sustainable. Vacuum cleaner robots, smartphones, and computer circuit boards also have to be disposed of at some point. Sustainable alternatives made from plants or fungi could help here.

Austrian researchers, for example, have developed a circuit board based on the shared tree fungus Shiny Lackporling, as they published in the journal Science Advances.

Circuit boards made from tree fungus

Circuit boards serve as carriers for electronic components and connect them to each other using so-called conductor tracks. The plate itself is made of a stable and electrically insulating material; plastic or silicon is usually used for this. Fungus, on the other hand, can create biodegradable electronic circuit boards that decompose themselves in a very short time, within several weeks.

This is made possible by so-called fungal mycelia. These are the root networks of fungi, which have vast networks of fibers underground. The mycelial skin is used for the circuit boards. The skin of the mushroom mycelia is both heat-resistant, robust and flexible.

Simple and resource-saving production

Production begins with beech shavings, wholemeal spelt flour, gypsum, and water—and with spores of the Shiny Lackporling. The research team from Johannes Kepler University Linz allowed the mycelium to grow on it. In the final step, the skin was peeled away from the mycelium, dried, pressed, and cut to the correct size. Conductor tracks can then be added, and electronic components can be attached as with conventional circuit boards.

According to the research team, circuit board production is more straightforward, requires less energy and water than conventional production, and does not require harmful chemicals. So far, this has produced simple and small printed circuit boards.

The researchers also use the fungal mycelia to make batteries. In such a battery, the mycelium of the shiny lacquer cell can consist of both the membrane between the poles and the cover.

Mushroom mycelium – a complex and adaptive network

In addition to the properties of the mycelial skin in electronics, the mycelium can also be attractive for science. Mushroom mycelium is a living, complex, and adaptable material that forms large networks. These networks, in turn, consist of elongated cells called hyphae. The hyphae absorb water and nutrients, which is how the fungus spreads in nature.

However, in most of the previously known application approaches, the fungi used die at the end of the process or are removed again. Researchers at the Swiss Federal Institute of Technology in Zurich also use the adaptable behaviour to develop self-healing and robust robot skin.

Living robot skin from the 3D printer

As the research team describes in Nature Materials, a three-dimensional grid is printed from a hydrogel using a 3D printer. The hydrogel is loaded with spores of the shiny lacquer pore. If you leave the framework at 23 degrees Celsius and a high relative humidity of 95 per cent for several days, the mycelium grows without the hydrogel drying out.

Within 20 days, the fungal mycelia colonise the printed grid, creating robust and regenerating skin. If this is cut or punctured, it will grow back together. The metabolic activity of the mycelia and the availability of nutrients are responsible for this.

Robot coated with mycelium.

The living robot skin of mycelium is soft, waterproof, regenerative and robust against mechanical influences. This means that the properties of the skin through the mycelium are comparable to some functions of biological animal skins.

The researchers carried out tests with a gripper arm and a ball robot covered with mycelium skin. The robots successfully completed underwater actions or were rolled over different surfaces.

Further research approaches and possible areas of application.

Both research approaches show that the use of fungal mycelia is still in its early stages. For example, complex circuit boards will be made from smoother mycelial skin in the future, and further research is also needed to keep the metabolic activity and, thus, the living robot skin alive in the long term.

But mushroom mycelium is also being used for research beyond electronics and robotics, for example, for sustainable insulation and building materials or for a durable leather alternative.

A bot with wheels moves along the surface. A star-shaped soft-bodied robot flexes its five legs, shifting with an unsteady shuffle.

While these basic robotic creations are powered by conventional electricity from a plug or battery, what makes these robots unique is that they are operated by a living organism: a king oyster mushroom.

A team of researchers from Cornell University has modified two types of robots by integrating the mushroom’s mycelium, or rootlike threads, into the hardware, enabling the robots to sense and respond to the environment by utilizing the fungus’s electrical signals and its sensitivity to light.

These robots represent the latest achievement in the field of biohybrid robotics, where scientists endeavor to combine biological, living materials such as plant and animal cells or insects with synthetic components to create entities that are partly living and partly engineered.

While biohybrid robots have not yet moved beyond the laboratory, researchers aspire to see robot jellyfish exploring the oceans, sperm-powered bots delivering fertility treatments, and cyborg cockroaches searching for survivors in the aftermath of an earthquake.

Robert Shepherd, a senior author of a study detailing the robots published in the journal Science Robotics on August 28, stated, “Mechanisms such as computing, understanding, and responsive action are accomplished in the biological world and in the artificial world created by humans, and most of the time, biology performs these tasks better than our artificial systems.”

“Biohybridization is an effort to identify components in the biological world that we can leverage, comprehend, and control to enhance the functionality of our artificial systems,” added Shepherd, who is a professor of mechanical and aerospace engineering at Cornell University and leads the institution’s Organic Robotics Lab.

The team initiated the process by cultivating king oyster mushrooms (Pleurotus eryngii) in the lab using a basic kit purchased online. The selection of this mushroom species was based on its ease and quickness of growth.

They grew the mushroom’s threadlike networks or mycelium, which, according to the study, can sense, communicate, and transport nutrients, functioning somewhat like neurons in a brain. (However, it is not entirely accurate to refer to the creations as “shroom bots.” The mushroom is the fruit of the fungi, while the robots are energized by the rootlike mycelium.)

The cultivation of the fungus in a petri dish took between 14 and 33 days to fully integrate with the robot’s framework, as per new research led by scientists at Cornell University.

Mycelium produces minor electrical signals and can be linked to electrodes.

Andrew Adamatzky, a professor of unconventional computing at the University of the West of England in Bristol who develops fungal computers, stated that it remains unclear how fungi generate electrical signals.

“No one knows for sure,” mentioned Adamatzky, who was not involved in the research but reviewed it before publication.

“Essentially, all living cells produce action-potential-like spikes, and fungi are no exception.”

The research team encountered difficulties in engineering a system capable of detecting and utilizing the small electrical signals from the mycelia to control the robot.

Anand Mishra, a postdoctoral research associate in Cornell’s Organic Robotics Lab and the lead author of the study, mentioned, “You have to ensure that your electrode makes contact in the correct position because the mycelia are very thin. There isn’t much biomass there. After that, you cultivate them, and as the mycelia start growing, they wrap around the electrode.”

Mishra developed an electrical interface that effectively captures the mycelia’s primary electrical activity, processes it, and converts it into digital information that can activate the robot’s actuators or moving components.

The robots were able to walk and roll in response to the electrical spikes produced by the mycelia, and when Mishra and his colleagues exposed the robots to ultraviolet light, they altered their movement and trajectory, demonstrating their ability to react to their surroundings.

“Mushrooms are not particularly fond of light,” Shepherd remarked. “Based on the variations in light intensities, you can elicit different functions from the robot. It will move more swiftly or distance itself from the light.”

“Exciting” progress

Victoria Webster-Wood, an associate professor at Carnegie Mellon University’s Biohybrid and Organic Robotics Group in Pittsburgh, mentioned the excitement surrounding further developments in biohybrid robotics beyond the utilization of human, animal, and insect tissues.

“Fungi may offer advantages over other biohybrid approaches in terms of the conditions required to sustain them,” Webster-Wood, who was not involved in the research, noted.

“If they are more resilient to environmental conditions, this could render them an exceptional candidate for applications in agriculture and marine monitoring or exploration.”

The study highlighted that fungi can be mass-cultivated and thrive in various environmental conditions.

The rolling robot was operated by the researchers without a tether connecting it to the electrical hardware — a notable accomplishment according to Webster-Wood.

Webster-Wood, via email, mentioned that truly tether-free biohybrid robots are a challenge in the field and it’s quite exciting to see them achieve this with the mycelium system.

Regarding real-world applications, Shepherd stated that fungi-controlled technology could be useful in agriculture.

Shepherd mentioned that in this case, light was used as the input, but in the future, it will be chemical. The potential for future robots could be to sense soil chemistry in row crops and decide when to add more fertilizer. This might help mitigate downstream effects of agriculture like harmful algal blooms, according to Shepherd.

Adamatzky emphasized the huge potential of fungi-controlled robots and fungal computing, mentioning that more than 30 sensing and computing devices using live fungi were produced in his lab. This included growing a self-healing skin for robots that can react to light and touch.

Adamatzky, via email, explained that when an adequate drivetrain is provided, the robot can, for example, monitor the health of ecological systems. The fungal controller would react to changes, such as air pollution, and guide the robot accordingly.

Mestre, who works on the social, ethical, and policy implications of emergent technologies, mentioned that if biohybrid robots become more sophisticated and are deployed in the ocean or another ecosystem, it could disrupt the habitat, challenging the traditional distinction between life and machine.

Mestre stated that if these robots are released in big numbers, it could be disruptive to the ecosystem. He also emphasized the importance of considering the ethical concerns as this research continues to develop.

Mushrooms have gained popularity as a vegan substitute for leather and are being used in high-end fashion and even in car manufacturing. Additionally, hallucinogenic varieties of mushrooms have been found to alleviate treatment-resistant depression.

Researchers at Johannes Kepler University in Linz, Austria, have found a significant use for fungi that could potentially help mitigate global warming.

The team, led by scientist Martin Kaltenbrunner, devised a way to use fungi as a biodegradable base material for electronics chips, as outlined in the journal Science Advances.

Kaltenbrunner, with a focus on sustainability, material science, and engineering, explored using sustainable materials in robotics in previous research.

In their latest research, the team looked at redesigning the substrate of electronic circuits utilizing a mushroom-based material to replace unrecyclable plastic polymers.

This mushroom, Ganoderma lucidum, was used for the experiment and has a history of promoting health and longevity in Asia. The team was particularly interested in the skin generated by this mushroom to cover its root-like appendage, called a mycelium.

When the skin was dried out and tested, it was discovered that it could endure temperatures of 200°C (390°F), and it acted as a good insulator and conductor. The skin could also easily hold circuit boards after being treated with metal and strengthened by the addition of copper, chromium, and gold.

Another positive characteristic of this remarkable fungi is its thickness, which is comparable to that of paper. Paper was considered as a potential substrate, but it was rejected due to its highly water-intensive and toxic chemical-soaked production process.

In contrast, the mushroom substrate could be bent up to 2,000 times without any damage and was so adaptable in shape that it surpassed the planar geometry challenges faced by engineers in chip design.

Andrew Adamatzky, a professor in unconventional computing at the University of the West of England, stated, “The prototypes produced are impressive and the results are groundbreaking,” in New Scientist.

Kaltenbrunner and his team anticipate that the mushroom-encased chip will be suitable for use in wearable, low-powered, and short-lived Bluetooth sensors for humidity and proximity, as well as in radio tags.

Moreover, the mycelium’s ability to repel moisture and UV light indicates that it could potentially endure for several hundred years. The research team has also proposed a completely new concept of batteries, having successfully used the mushroom skins as battery separators and casings.

Even more encouraging, the production of these mushrooms has minimal impact on the environment — in fact, the more CO2 available for their production, the better. The team effortlessly grew and harvested mature mycelium on beechwood in just four weeks.

Furthermore, when these devices reach the end of their lifespan, they can biodegrade quietly in any soil and disappear in less than two weeks, presenting the kind of solution that engineers need to adopt in order to counter the unsustainable electronic consumption threatening the world.

Introduction

In a world affected by climate change and extensive waste production, environmental impact must be a primary consideration in technological innovations. Disposable technology, in particular, represents an increasingly large portion of our waste, accumulating over 100,000 tons per day. End-of-life consumer electronics, which are often difficult to recycle due to diverse product designs and material compositions, are typically discarded since they are cheaply produced. In addition, the unsustainable use of rare and often toxic materials poses an environmental threat when inadequately treated or landfilled.

Designs for easily recyclable devices, the use of low-cost and renewable materials, and the implementation of biodegradable or transient systems are promising approaches toward technologies with a closed life cycle, opening up new opportunities in various fields from medicine and environmental monitoring to security and intelligence applications.

Recent advancements in robotics focusing on safe human-machine interaction, swarm robotics, and untethered autonomous operation are frequently inspired by the diversity found in nature. The intricacy observed in nature motivates scientists from various disciplines to develop soft and lightweight forms of robots that aim to replicate or mimic the graceful movements of animals or their efficient energy management.

In the future, the increased integration of such soft robots into our daily lives poses, akin to consumer electronics, environmental concerns at the end of their life cycle. Once again, we can derive inspiration from nature and design our creations in a sustainable manner, mitigating the issues associated with current technology. Unlike standardized industrial robots, which are already incorporated into recycling loops, bioinspired robotics will find diverse ecological applications in various niches.

Examples range from soft healthcare machines that assist elderly individuals in their daily activities to robots that harvest produce and then decompose as compost for the next season’s plants. Ongoing demonstrations of transient behavior include elastic pneumatic actuators, in vivo-operating millibots for wound patching, robot swarms for drug delivery, and small grippers controlled by engineered muscle tissues.

These developments benefit from extensive research efforts towards bioresorbable electronic devices, primarily explored in the biomedical sector, and sustainable energy storage technology, aiming to address environmental concerns associated with the growing demand for energy in mobile devices. The future challenge for autonomous robots will be the efficient integration of actuators, sensors, computation, and energy into a single robot, requiring novel concepts and eco-friendly solutions. Success can only be achieved by bringing together material scientists, chemists, engineers, biologists, computer scientists, and roboticists.

Here, we present materials, manufacturing methods, and design strategies for eco-friendly bioinspired robots and their components. Our focus is on sustainable device concepts, non-toxic, and low-cost production processes, and environmentally safe materials that are either biodegradable or sourced from renewable resources, all of which address the current pressing needs. The review begins with an exploration of sustainability and summarizes various approaches that enable technology with reduced environmental impact.

Turning our attention to soft and lightweight forms of robotics, we then compare biodegradable polymers—from elastomers to bioplastics—and regenerative resources for the primary robotic body. In each component of typical autonomous robots, we examine environmentally friendly sensors, computation, and control tools, and introduce promising options for energy harvesters and storage systems, including solar- and biofuel cells, as well as batteries. Lastly, we showcase a selection of current soft robotic demonstrations that utilize economical material approaches and degrade with a positive impact on the surroundings.

Sustainable Approaches for Soft Robotics

The main scientific inquiries into sustainable materials development for robotics revolve around two questions. First, can we use new materials and resources that contribute to a more sustainable future? Second, how can we utilize or modify existing materials to reduce their ecological footprint on the environment?

Addressing the first question involves the development of high-performance materials with increased durability, materials sourced from renewable sources, or biodegradable ones, all aiming to conserve valuable resources or minimize waste. Similar objectives apply to solutions addressing the second question, which focus on fabrication processes, recycling, and product designs. Sustainability in robotics encompasses numerous facets, approaches, and solutions, which we delve into in this section, including renewable resources, recycling, and biodegradability.

Renewable Resources

Unlike finite resources such as fossil fuels, nuclear fuels, and rare earth metals, renewable materials are either perpetually available or naturally replenished within reasonable timeframes. In an ideal sustainable scenario, the consumption rates of material/energy equal the regeneration rate of the resources. Autonomous robotics stand to benefit from renewable resources more than other technologies, by harnessing energy from solar power or tidal waves and by replacing damaged body parts with spare parts that naturally regenerate.

Solar power, a long-standing standard for space exploration robots, offers an inexhaustible energy supply that can be stored in a robot’s battery to provide consistent power over an extended period. The smaller and lighter a robot is, the more efficient it becomes to utilize solar power over fuel energy, as robots only need to carry collectors, not the fuel itself. For instance, extremely lightweight solar panels can deliver substantial power (23 W g−1) without adding considerable weight to the robot.

Rather than using fossil-based plastics, the robotic body can be constructed from plant-based materials. Green composite materials show promise as suitable candidates for sturdy yet lightweight components, not only for robots but also for mobile machinery in general. In the context of electric cars, lightweight natural fiber composites with adequate mechanical properties could replace dense synthetic materials for both interior and exterior components, helping to offset the increasing weight of batteries.

To cater to the growing interaction between machines and humans, elastomers derived from biomaterials can be used to create soft grippers or (robotic) soft electronic skins (e-skins) that mimic biological designs. Carbonized biomass can be employed as an electron conductive alternative to metals in many electronic components, or it can participate in the electrochemical reactions of batteries and supercapacitors.

However, the use of renewable materials primarily addresses resource issues rather than waste issues. For instance, vulcanized natural rubber, despite being naturally derived, does not degrade within a reasonable timeframe and necessitates waste treatment and recycling. Therefore, renewability, biodegradability, and recycling must be collectively optimized to yield a sustainable technology with a beneficial impact on resources and waste.

Recycling

For technologies that must meet high performance benchmarks—such as complementary metal-oxide-semiconductor (CMOS) chips or Bluetooth communication—finding renewable or biodegradable alternatives remains challenging. Thus, recycling emerges as a viable approach toward the more sustainable use of technology. It is important to view recycling as the process of transforming waste into a valuable (similar) product. Recycling also encompasses the generation of energy through waste combustion, although this is only sustainable to a certain extent, as it consumes resources and elevates CO2 emissions.

In general, whether it’s material, device, or robot recycling, the decision is often driven by economic considerations: a product is more likely to be recycled if the cost of recycling is lower than the cost of manufacturing a new one.

As a result, an effective recycling process must be economically viable, easily achievable technologically, integrated into closed production-recycling loops, focused on valuable materials, and requiring minimal energy. An example of efficient recycling is lead-acid batteries (such as car batteries). Due to their standardized simple design, these batteries can be easily taken apart and recycled. When technicians replace the batteries, they close the life-cycle loop by sending the worn-out batteries back to the manufacturers.

Recycling other electronic waste (e-waste) is often challenging and not easily achievable due to the varying architecture and material composition of integrated circuits, Li-batteries, or displays. To reduce recycling costs, e-waste is sometimes sent to developing countries like Ghana, where improper e-waste processing endangers workers and residents.

To make robotics sustainable, recycling must be considered during the design phase. A successful recycling plan necessitates the easy separability of individual robotic materials to facilitate straightforward reuse, exchange, and upgrading of robots. While this is more feasible for traditional robots, as they often consist of standardized electronic parts and actuators, it can be difficult for soft robots, which employ various actuation principles and materials. However, soft robots benefit from less complex material arrangements.

For instance, pneumatically driven soft robots have combined actuators and bodies. As a result, the complexity of recycling an entire robotic body with many actuators (comprised of various components themselves) is reduced to recycling a single material.

Similarly, the less stringent requirements of control feedback allow for e-skins with reduced material complexity. A beneficial approach is to incorporate self-healing materials or concepts for soft robots that autonomously restore materials functionality. Tan and colleagues developed a stretchable optoelectronic material for stretchable electronics and soft robotics with light emission and feedback sensing, which independently self-heals after being punctured.

Another sustainable approach involves using fewer materials in the design. Autonomous robots benefit twofold from lightweight materials/component designs, aiming to first reduce weight and increase operation time, and second minimize environmental impact by decreasing the total amount of waste. Ultimately, zero waste robotics could be achieved with fully biodegradable materials.

Biodegradable materials are a promising material class for sustainable technology. In the ideal scenario, a material breaks down into smaller, environmentally friendly components that are metabolized by bacteria or enzymes at timescales comparable to typical waste processing. Moreover, the degradation process should start at the end-of-life phase of a device, triggered and occurring at a controlled rate and under feasible environmental conditions. The concept of biodegradability is not clearly defined and handled consistently in the literature, particularly concerning multicomponent/material devices.

For biodegradable electronics, not all components may be biodegradable, or they may degrade at different rates. Bao and colleagues distinguish between materials with transient behavior (type I) that disintegrate into sufficiently small components and biodegradable materials (type II) that undergo complete chemical degradation into tiny molecules.

Transient electronics, made from type I materials, play a significant role in the biomedical sector. Implantable or ingestible devices are designed to remain in our bodies, monitoring cardiac pressure, glucose levels, or neural activities. The degradation of these devices must be achievable under physiological conditions to create truly bioresorbable devices. Therefore, the lifetime of all materials should be limited to timescales comparable to the healing of human tissue or regeneration processes, and each degradation product must be noncytotoxic.

Such material design also holds promise for microbots operating inside the body, for wound treatment or drug delivery applications. Outside the body, biodegradable materials enable secure systems that disappear after their operation, preventing plagiarism, espionage, or unauthorized acquisition of critical technology.

Biodegradable robotics and electronics (type II) require the complete metabolization of all constituents. It is not enough for materials to break down into smaller units; they must be converted into biomass or gases by microorganisms. Additionally, materials that degrade into bio-derived small molecules offer intrinsic biocompatibility and recyclability, returning energy back to nature. This technology may ultimately provide solutions to critical e-waste issues while transforming conventional robotics into creative solutions that encompass the entire technology life cycle.

To ensure the correct degradation of materials, it is crucial to accurately report the application areas, operational environments, and degradation timescales for type I or II technology. Implanted devices should degrade under conditions similar to our body’s environment, produce harvesting robots must decompose in organic waste and compost, while maritime fish robots need materials that disintegrate in seawater.

Immersing a material into an unsuitable environment might not result in any degradation, even if it is labeled biodegradable. This misunderstanding is unfortunately common in reports of biodegradable materials, as illustrated by Bagheri and colleagues.

For their study on degradation, Bagheri and co-workers immersed typical biodegradable polymers like polylactic acid (PLA), polycaprolactone (PCL), and poly(3-hydroxybutyrate) (P3HB) in seawater. Surprisingly, they discovered that these polymers hardly degrade over 400 days, with a mass loss of less than 10%. The same holds true for the elastomer Ecoflex used in the soft robotics community. Although this polymer is 100% fossil-based, it fully decomposes in approximately 80 days under industrial composting conditions.

Cellulose, for instance, requires about 50 days under the same conditions. In seawater, factors such as temperature, microorganisms, and oxygen availability differ significantly from those in compost, leading to a much longer degradation time for Ecoflex.

While there are also standards for biodegradation in seawater, the most common standards that certify biodegradable polymers, particularly in packaging, target degradation in industrial composting facilities. The ISO 17088 norm, effective since 2008, is the globally applicable standard based on the European EN13432 and American ASTM 6400-04 standards. In essence, biodegradation tests monitor the CO2 evolution of polymer/compost mixtures under optimum humidity and oxygen conditions at 58 °C, with specified pass levels.

In situations where industrial composting is not feasible, biodegradable materials are necessary to disintegrate in less controlled environments. For instance, tech waste disposed of through household composts or in nature needs to vanish under milder conditions, yet at equally rapid rates.

For biodegradable materials used in electronics or robotics, additional declarations should indicate that the robot, once its purpose is fulfilled and it reaches the end of its life cycle, can simply be discarded without consideration for environmental conditions or be left at the disposal site. Therefore , advancing materials that enable individual-based waste management requires research, standards, and specifications.

A wheeled robot traverses the ground. A soft-bodied robotic star shifts its five legs, moving in a somewhat clumsy manner.

These basic robotic creations would be considered ordinary if not for one distinguishing feature: they are controlled by a living organism—a king oyster mushroom.

By integrating the mushroom’s mycelium, or rootlike filaments, into the robot’s design, researchers from Cornell University have created two types of robots that perceive and react to their surroundings by utilizing electrical signals generated by the fungus and its light sensitivity.

These robots represent the latest achievement in the field of biohybrid robotics, where scientists aim to merge biological materials, such as plant and animal cells or insects, with artificial components to create entities that are partially alive and partially engineered.

Although biohybrid robots have not yet left the laboratory, researchers are optimistic that future applications could include robot jellyfish exploring the oceans, sperm-driven robots delivering fertility treatments, and cyborg cockroaches searching for survivors after earthquakes.

“Biological mechanisms, including computing, comprehension, and actions in response, exist in nature, often outperforming the artificial systems developed by humans,” stated Robert Shepherd, a senior author of a study about the robots published on August 28 in the journal Science Robotics.

“Biohybridization endeavors to identify biological components that we can utilize, comprehend, and control to enhance the performance of our artificial systems,” added Shepherd, a professor of mechanical and aerospace engineering at Cornell University and head of the school’s Organic Robotics Lab.

A combination of fungus and machinery

The research team began by cultivating king oyster mushrooms (Pleurotus eryngii) in the lab using a basic kit purchased online. They selected this mushroom species because it is simple and quick to grow.

They grew the mushroom’s threadlike structures, or mycelium, which can develop networks capable of sensing, communicating, and transporting nutrients—similar in function to neurons in a brain. (It’s important to note that referring to these as shroom bots isn’t entirely correct, as the robots derive their power from the rootlike mycelium, not the mushroom itself.)

Mycelium emits small electrical signals and can be linked to electrodes.

Andrew Adamatzky, a professor specializing in unconventional computing at the University of the West of England in Bristol who constructs fungal computers, stated that the exact mechanism by which fungi generate electrical signals remains uncertain.

“Currently, nobody knows for certain,” said Adamatzky, who did not participate in the study but reviewed it prior to publication.

“Basically, all living cells generate action-potential-like spikes, and fungi are no different.”

The research team encountered difficulties in creating a system that could identify and utilize the faint electrical signals from the mycelia to control the robot.

“It’s essential to ensure that your electrode is positioned correctly because the mycelia are extremely fine. There is minimal biomass present,” explained lead author Anand Mishra, a postdoctoral research associate in Cornell’s Organic Robotics Lab. “Afterward, you culture them, and as the mycelia begin to grow, they wrap around the electrode.”

Mishra developed an electrical interface that effectively reads the mycelia’s raw electrical activity, processes it, and converts it into digital signals capable of activating the robot’s actuators or moving parts.

The robots demonstrated the ability to walk and roll in response to electrical spikes generated by the mycelia, and when stimulated with ultraviolet light, they altered their gait and trajectory, indicating that they could react to their environment.

“Mushrooms tend to shy away from light,” Shepherd remarked. “By varying light intensities, you can induce different functions in the robot. It might move faster or steer away from the light.”

‘Exciting’ progress

The advancements in biohybrid robotics that extend beyond human, animal, and insect tissues are exhilarating, noted Victoria Webster-Wood, an associate professor at Carnegie Mellon University’s Biohybrid and Organic Robotics Group in Pittsburgh.

“Fungi may offer advantages over other biohybrid strategies regarding the environmental conditions needed for their survival,” stated Webster-Wood, who was not part of the research.

“If they can withstand environmental variations, it could make them an excellent choice for biohybrid robots used in agriculture, marine monitoring, or exploratory purposes.”

The research highlighted that fungi can be grown in significant volumes and can prosper in a variety of environments.

The team operated the rolling robot without a tether linking it to the electrical components — a task that Webster-Wood emphasized as particularly significant.

“Completely tetherless biohybrid robots pose a challenge in this field,” she mentioned in an email, “and witnessing their accomplishment with the mycelium system is extremely thrilling.”

Fungi-managed technology could find uses in agriculture, as noted by Shepherd.

“In this scenario, we utilized light as the stimulus, but in the future, it will likely be chemical. The future possibilities for robots might include detecting soil chemistry in crop rows and determining when to apply additional fertilizer, potentially alleviating the negative downstream impacts of agriculture such as harmful algal blooms,” he explained to the Cornell Chronicle.

According to Adamatzky, fungi-controlled robots, and fungal computing in a broader sense, hold significant promise.

He stated that his laboratory has developed over 30 devices for sensing and computing using live fungi, including creating a self-repairing skin for robots that can respond to both light and touch.

“With a suitable drivetrain (transmission system) in place, the robot could, for instance, assess the condition of ecological systems. The fungal controller would respond to variations like air pollution and direct the robot accordingly,” Adamatzky wrote in an email.

“The emergence of yet another fungal device — a robotic controller — excitingly showcases the extraordinary potential of fungi.”

Rafael Mestre, a lecturer at the University of Southampton’s School of Electronics and Computer Science in the UK, who focuses on the social, ethical, and policy implications of emerging technologies, expressed that if biohybrid robots become increasingly advanced and are introduced into oceanic or other ecosystems, it could disrupt the environment, challenging the conventional boundaries between living organisms and machines.

“You are introducing these entities into the food web of an ecosystem where they may not belong,” Mestre remarked, who was not part of the recent study. “If they are released in significant quantities, it could be disruptive. At this time, I don’t perceive strong ethical concerns surrounding this specific research… but as it continues to evolve, it is essential to contemplate the consequences of releasing this into the wild.”

November 6, 2024
AI Uses Driving Data to Warn of Low Blood Sugar
Diabetic hypoglycemia occurs when someone with diabetes doesn’t have enough sugar (glucose) in his or her blood. Glucose is the main source of fuel for the body and brain, so you can’t function well if you don’t have enough.

For many people, low blood sugar (hypoglycemia) is a blood sugar level below 70 milligrams per deciliter (mg/dL) or 3.9 millimoles per liter (mmol/L). But your numbers might be different. Ask your health care provider about the appropriate range to keep your blood sugar (target range).
- Neurogenic or neuroglycopenic symptoms of hypoglycemia may be categorized as follows:
- Neurogenic (adrenergic) (sympathoadrenal activation)symptoms: Sweating, shakiness, tachycardia, anxiety, and a sensation of hunger
- Neuroglycopenic symptoms: Weakness, tiredness, or dizziness; inappropriate behavior (sometimes mistaken for inebration); difficulty with concentration; confusion; blurred vision; and, in extreme cases, coma and death
Please pay attention to the early warning signs of hypoglycemia and treat low blood sugar as soon as possible. You can raise your blood sugar quickly by eating or drinking a simple sugar source, such as glucose tablets, hard candy or fruit juice. Could you tell family and friends what symptoms to look for and what to do if you’re not able to treat the condition yourself?

Before people with diabetes experience hypoglycemia while driving, they could be warned by an AI in the future. Researchers from Munich and Switzerland have successfully tested such an application.

To warn drivers who have diabetes in good time about hypoglycemia in the future, researchers are working on a system that uses artificial intelligence (AI). Scientists from the Ludwig Maximilian University of Munich (LMU) and researchers from ETH Zurich, the Berner Inselspital and the University of St. Gallen are involved.

Test drives before and after induced hypoglycemia

The researchers tested their AI model in a large-scale driving test on a military site in Thun, Switzerland. The driving patients were each accompanied by a driving instructor next to them and two or three medical professionals in the back seat.

After initially driving with normal blood sugar levels, they administered continuously insulin to the driver so that the blood sugar level became lower and lower. The corresponding data was recorded to develop an AI model.

Analysis of the driving behavior of hypoglycemic patients

Simon Schallmoser, a doctoral student at the LMU, is writing a doctoral thesis on this topic and has evaluated the driving data for his AImodel, as well as the head and eye movements recorded by camera of those with artificial hypoglycemia who were behind the wheel.

When a person experiences hypoglycemia, their movements change. To be more precise, the look and position of the head become a little more monotonous. People with hypoglycemia tend to look in the direction of gaze for longer, and when they change their direction of gaze , it happens more quickly.

They are no longer quite as forward-looking, and this can also be measured using the car’s driving signals, explains Schallmoser: “For example, we noticed that patients with low blood sugar levels make fewer small corrections when steering, which we know from driving a car, but rather change the direction of travel very abruptly.”

Tests in real road traffic are still pending

The AI application was tested on a test track at Anairfield, where driving in city traffic, on country roads and on the motorway was simulated with 30 patients. That’s enough for the AI application to be meaningful, says Simon Schallmoser. However , according to Schallmoser, new experiments would have to be carried out before it could actually be used in real road traffic, as the test route only had limited significance for real road traffic.

The researcher explains that further studies will be necessary before it is ready for the market. However, the first tests as to whether artificial intelligence detects hypoglycemia have already been very promising.“We trained the model on patients and then tested it on other patients in the same study,” says Simon Schallmoser. “In machine learning, we talk about the fact that training and test data sets must not match; that is, the patients must not overlap. That’s how we tested it, and it worked very well. ”

Other possible uses are conceivable

Further tests are required, as well as cooperation with interested car manufacturers to install such systems in vehicles. This software upgrade is for well-equipped, modern cars, as the camera to detect drowsiness is already on board.

The question remains whether the AI application could be used for other purposes, such as detecting hypoglycemia and perhaps alcohol consumption. These tests are still pending. A large supplier was already involved in the test drives.

Detecting Diabetic Eye Disease Through AI Learning

Researchers at the Google Brain initiative have utilized “deep learning” methods to develop a self-optimizing algorithm that can analyze large quantities of fundus photographs and automatically identify diabetic retinopathy (DR) and diabetic macular edema (DME) with a high level of precision.

When the performance of the screening algorithm was evaluated by the researchers using 2 groups of images (N = 11,711), it demonstrated a sensitivity of 96.1% and 97.5% and a specificity of 93.9% for DR and DME, respectively.1

Peter A. Karth, MD, MBA, a vitreoretinal subspecialist in Eugene, Ore., and at Stanford University, who is a consultant to the Google Brain project, acknowledged the achievement, stating, “It’s a real accomplishment that Google was able to get high sensitivity and specificity at the same time—meaning that not only is this algorithm missing very few people who have disease, but it is also unlikely to overdiagnose disease.”

The algorithm operates based on deep machine learning, which is a form of artificial intelligence (AI) technology where a neural network “learns” to carry out a task through repetition and self-correction.

In this instance, the authors noted that the computerized algorithm was trained using 128,175 human-graded fundus images showing different levels of diabetic retinal disease. The authors explained, “Then, for each image, the severity grade given by the [algorithm] is compared with the known grade from the training set, and parameters … are then modified slightly to decrease the error on that image.” They added, “This process is repeated for every image in the training set many times over, and the [algorithm] ‘ learns’ how to accurately compute the diabetic retinopathy severity from the pixel intensities of the image for all images in the training set.”

According to Dr. Karth, the algorithm is effective despite not being designed to specifically search for the lesion-based features that a human would look for on fundus images. He stated, “What’s so exciting with deep learning is that we’re not actually yet sure what the system is looking at. All we know is that it’s arriving at a correct diagnosis as often as ophthalmologists are.”

Ehsan Rahimy, MD, a Google Brain consultant and vitreoretinal subspecialist in practice at the Palo Alto Medical Foundation, in Palo Alto, Calif., expressed similar sentiments, stating, “We don’t entirely understand the path that the system is taking. may very well be seeing the same things we’re seeing, like microaneurysms, hemorrhages, or neovascularization.”

AI Will Not Replace Doctors’ Intelligence

Dr. Karth and Dr. Rahimy highlighted that although additional work is required before the algorithm is ready for clinical use, the ultimate objective is to enhance access to and reduce the cost of screening and treatment for diabetic eye disease, particularly in under-resourced environments.

Dr. Rahimy emphasized, “Anytime you talk about machine learning in medicine, the knee-jerk reaction is to worry that doctors are being replaced. But this is not going to replace doctors. In fact, it’s going to increase the flow of patients with real disease who needs real treatments.”

Dr. Karth added, “This is an important first step toward dramatically lowering the cost of screening for diabetic retinopathy and, therefore, dramatically increasing the number of people who are screened.”

The Role of AI in Healthcare

AI has emerged as a revolutionary tool in healthcare, and with its ability to process extensive data, it has the potential to transform the accuracy and effectiveness of diagnostics and predictive decision-making. While AI offers numerous benefits and possibilities for diabetes research, diagnosis, and prognosis, it also comes with limitations.

Understanding AI in Healthcare

Artificial intelligence involves the simulation of human intelligence in machines programmed to think and learn like humans. In the healthcare sector, AI technologies, such as machine learning and deep learning, have made significant strides due to enhanced computer speed and increased computational resources.

Machine learning entails training algorithms to recognize patterns and make data-based predictions, commonly known as predictive analytics. On the other hand, deep learning utilizes neural networks to process intricate information and extract meaningful insights. These AI technologies enable healthcare professionals to analyze extensive datasets and derive valuable conclusions to enhance patient care.

AI’s efficacy lies in its ability to identify diabetes-related complications using comprehensive datasets and advanced algorithms.

How AI Can Enhance Diabetes Care

Accurate and timely diagnosis and treatment are crucial for effective diabetes management. AI’s effectiveness stems from its capability to identify diabetes-related complications using extensive datasets and advanced algorithms.

For instance, AI-based medical devices have been authorized for automated retinal screening to identify diabetic retinopathy (DR) from fundus images. The IDx-DR device, approved by the FDA for DR diagnosis, can provide a diagnosis without requiring professional judgment from an ophthalmologist. Its use has been especially beneficial for rural communities with limited access to specialized healthcare professionals.

AI, with its capability to fine-tune insulin doses and enhance decision-making processes, can significantly aid in clinical treatment. Systems like Advisor Pro, which employs AI algorithms to analyze continuous glucose monitoring (CGM) and self-monitoring blood glucose (SMBG) ) data, can facilitate remote insulin dose adjustments. This technology empowers healthcare professionals to make informed decisions to support their patients’ self-care.

AI can also help with risk stratification, allowing healthcare professionals to identify high-risk individuals and offer targeted interventions. Machine learning algorithms can assess patient data, including medical history, lifestyle factors, and genetic markers, to predict the likelihood of developing diabetes or its complications. This information can guide preventive measures and personalized treatment plans.

It is crucial to understand how AI reaches its conclusions to gain trust and acceptance from healthcare professionals and patients.

Constraints and difficulties of AI

While AI holds significant promise in diabetes research and management, it is important to recognize its limitations and challenges. One primary concern is the interpretability and explainability of AI algorithms. Unlike traditional statistical models, AI algorithms can be perceived as “black boxes” due to their complex decision-making processes. It is critical to understand how AI arrives at its conclusions to gain trust and acceptance from healthcare professionals and patients.

Addressing the challenges of AI in diabetes management, such as the requirement for high-quality, diverse, and well-annotated datasets, necessitates a collaborative effort. AI heavily relies on training data to learn patterns and make accurate predictions. However, data bias and limited access to comprehensive datasets can impede the performance and generalizability of AI models. Therefore, it is crucial for researchers, healthcare institutions, and regulatory bodies to collaborate to ensure robust and representative data availability.

Furthermore, regulatory frameworks must keep pace with the rapid advancements in AI technology. Clear guidelines and standards are needed to ensure safe and ethical use of AI in healthcare. Other considerations such as data privacy, security, and patient confidentiality are also crucial to build public trust in AI-driven healthcare solutions.

As technology and medical science progress, the accuracy and predictive performance of AI algorithms will also improve.

Looking ahead with AI

Despite the challenges, ongoing research and innovation in AI hold significant promise for diabetes care. As technology and medical science advance, the accuracy and predictive performance of AI algorithms will also improve.

Organized data and ample computational capacity will optimize AI’s forecasting capabilities, leading to more accurate disease prediction models for diabetes. This progress instills hope for a future where AI can significantly improve patient outcomes and transform diabetes management.

As we look to the future, collaboration between researchers, healthcare professionals, and technology experts will be crucial in harnessing AI’s full potential in diabetes management. By overcoming challenges and leveraging AI’s power, we can pave the way for a future where diabetes is better understood , managed, and ultimately prevented.

With the rise of digitalization, we have observed diabetes management expanding beyond commonly used devices to smartphone apps.

Innovations in diabetes management technologies can offer more effective and manageable treatment options, ultimately transforming the landscape of diabetes care. With the rise of digitalization, we have seen diabetes management expanding beyond commonly used devices to smartphone applications – apps for short.

An app is self-contained software crafted for a mobile device – smartphones, tablets, laptops, or desktop computers – that enables users to carry out specific tasks. Apps are particularly convenient when used on mobile devices. They can be utilized offline, providing access to information and features even without an internet connection. Mobile apps can also send notifications to users, providing real-time updates.

Popular features in diabetes mobile apps

A range of features found in diabetes mobile apps can make diabetes management more convenient. These features enable users to record insulin, physical activity, and carbohydrate intake, and monitor crucial health data, all while gathering data directly from continuous glucose monitors (CGMs). Some even offer distinct features, such as low blood glucose alerts.

Alternative types of applications provide features for diabetes. They connect various blood glucose meters (BGMs), continuous glucose monitors (CGMs), and insulin pumps to create detailed charts of blood glucose levels, insulin managing extensive data. Users can create personalized care plans in collaboration with their diabetes care team. Additionally, these apps are widely available on different platforms, ensuring accessibility for users regardless of location or device.

There are manageable potential risks, but the benefits and conveniences offered by diabetes apps outweigh the drawbacks, making them valuable in diabetes care.

While diabetes apps offer many advantages, they also come with potential drawbacks. These include the need for frequent updates as the app evolves and a necessity for greater regulation to prevent bugs or security risks. However, it’s crucial to note that these potential risks can be managed, and the benefits and conveniences offered make diabetes apps valuable in diabetes care.

Selecting a diabetes management app should involve following good practices. Trying out multiple apps before deciding on the most suitable one is recommended. Consider your preferences, goals, and the need for a personalized diabetes management plan. Healthcare providers can often assist their patients in understanding how to use an app, interpreting data, and providing guidance on any limitations, ensuring an informed decision.

Recent technological advancements in diabetes management have made it easier to synchronize automated insulin delivery systems (AID) and continuous glucose monitors (CGMs) with an app. AID systems combine an insulin pump and CGM to help people with diabetes monitor their blood glucose levels. intelligent algorithm links the two devices, enabling them to exchange data. AIDs can improve glycemic control through real-time responses, ultimately reducing the burden of manual insulin dosing.

For diabetes management, there are electronic platforms known as Diabetes Management Platforms (DMPs) which can aid people with diabetes. DMPs collect data from diabetes devices (BGM, CGM, or insulin pump) through a synced mobile app, and this data can also be accessed online for manual logging.

Diabetes management platforms utilize AI and CGMs to provide personalized management strategies by predicting blood glucose levels and optimizing insulin dosages. They can also address accessibility issues by ensuring the latest diabetes technology is available from the time of diagnosis. DMPs using AI incorporate an algorithm-powered dashboard that consolidates data from different diabetes devices and presents it in a user-friendly manner for healthcare providers, enhancing diabetes care and management.

The future of DMPs looks promising, with continuous technological advancements offering improved app functionalities. By transitioning from traditional pencil-logbook methods to sophisticated data logging and analysis, DMPs have the potential to revolutionize diabetes management. Furthermore, these platform advancements can support healthcare providers in guiding their patients toward practical diabetes management tools.

The use of artificial intelligence (AI) in diabetes care has been focused on early intervention and treatment management. Notably, this usage has expanded to predict an individual’s risk of developing type 2 diabetes. A scoping review of 40 studies by Mohsen et al. shows that while most studies used single AI models, those that used multiple types of data were more effective. However, creating and determining the performance of these multi-faced models can be challenging due to the many factors involved in diabetes.

For both single and multi-faced models, concerns exist regarding bias due to the lack of external validations and representation of race, age, and gender in training data. Developing new technologies, especially for entrepreneurs and innovators, in the areas of data quality and evaluation standardization is crucial. Collaboration among providers, entrepreneurs, and researchers must be prioritized to ensure that AI in diabetes care provides quality and equitable patient care.

Introduction

Given the urgent need to address the increasing incidence and prevalence of diabetes on a global scale, promising new applications of artificial intelligence (AI) for this chronic disease have emerged. These applications encompass the development of predictive models, risk stratification, evaluation of novel risk predictors, and therapeutic management.

So far, most FDA-approved AI tools have been designed for early intervention and treatment management. Several of these tools are currently used in clinical diabetes care. For early intervention, in 2018, the FDA approved the autonomous AI system Digital Diagnostics, which demonstrated high diagnostic accuracy in recognizing diabetes retinopathy in retinal screening images.

The Guardian Connect System, which utilizes AI technology, was approved by the FDA in the same year to analyze biomedical data and forecast a hypoglycemic attack one hour in advance. Subsequently, the FDA has sanctioned AI technologies aiding in optimizing insulin dosing and therapy for patients .

AI is now being used to anticipate an individual’s risk of developing type 2 diabetes (T2DM) and potential complications, aside from intervention and treatment. Recognizing high-risk individuals and customizing prevention strategies and targeted treatments could delay or prevent the onset of diabetes and future health complications.

A scoping review by Mohsen et al. examined 40 studies that looked into AI-based models for diabetes risk prediction. Most studies gauged the performance of the area using the area under the curve (AUC) metric, a common metric in machine learning algorithms. AUC value of 1 denotes a perfect model.

The majority of these models were classical machine learning models with electronic health records as the primary data source. Although a limited number of studies (n = 10) employed multimodal approaches, they outperformed unimodal models (n = 30).

For instance, one multimodal approach found that a model integrating genomic, metabolomic, and clinical risk factors was superior in predicting T2DM (AUC of 0.96) compared to genomics-only (AUC of 0.586) and clinical-only models (AUC of 0.798).

However, developing multimodal models is highly time-consuming, making it challenging to scale such models easily. Moreover, integrating data sources can complicate the understanding of interactions among modalities and the rationale behind predictions, resulting in a scarcity of multimodal AI models for T2DM.

Although the review by Mohsen et al. suggests promising AI technologies for T2DM risk prediction, the findings should be approached cautiously. Determining the best-performing model is challenging due to the influence of various input risk predictors for diabetes.

For example, the XGBoost algorithm was used in three unimodal studies but yielded widely disparate AUC values (0.91, 0.83, and 0.679) due to variations in risk predictors and datasets.

Moreover, there are concerns regarding bias stemming from the demographic representation across models, with many showing imbalanced gender, ethnicity, and age. Most studies did not evaluate the algorithm’s performance across different demographic groups, hence perpetuating existing health inequities for already at-risk populations.

To ensure demographic representation in datasets, it is necessary to implement policies that require mandatory representation criteria for approval and adoption. It is important to integrate appropriate evaluation metrics, such as using Quality Assessment of Diagnostic Accuracy Studies (QUADAS) AI frameworks to evaluate a model’s risk of bias. External validation is also crucial to ensure the models’ generalizability beyond specific training datasets.

The QUADAS AI tool is a tool based on evidence designed to evaluate bias risk—related to patient selection, diagnostic test interpretation, and choice of reference standard—and applicability—generalizability of a study’s findings to the intended population—of diagnostic accuracy studies AI Adopting a comprehensive approach will ensure the use of fair and impartial AI models in order to prevent worsening existing health discrepancies.

Coming Soon

AI tools in diabetes care, specifically those trained with a multimodal approach, have promising applications in risk prediction. However, as unimodal approaches are still more prevalent, there exists untapped potential in employing more precise tools that match the standards of clinical care patients deserve. Innovative solutions are required on two fronts—data quality and standardized assessment metrics.

To build accurate tools, it is essential to have comprehensive and diverse datasets to train models. Especially as health data continues to be gathered to create robust datasets, there is a need to organize and structure the data for potential compatibility and interoperability when developing multimodal algorithms Universal evaluation protocols are also required to minimize the perpetuation of health inequalities.

The widespread and rapid adoption of AI in healthcare cannot happen until the issues related to data quality and bias are addressed—making these two aspects prime areas of development for innovations and new technologies from the private sector. Solutions that foster collaboration and transparency on these two fronts could draw inspiration from structures in other AI, such as open-source fields platforms, ethical review processes, and enforcement of bias testing in order to uphold a higher standard of practice.

In order to ensure that patient care is the primary focus of innovative AI tools in diabetes care, solutions must stem from collaborative efforts with all stakeholders—clinicians, researchers, policymakers, and entrepreneurs—as we continue to drive progress in the field of AI and diabetes.

Artificial intelligence and diabetes are two topics that are dear to me. This is why, in celebration of World Diabetes Day, I have chosen to share the numerous fascinating ways AI is assisting the medical field in the battle against the disease.

Whether you have diabetes or not, I am confident you will appreciate the innovative capabilities of humankind.

Acknowledging World Diabetes Day

On Saturday, November 14th, the world turned its attention to World Diabetes Day: an annual global campaign aimed at raising awareness about diabetes. The International Diabetology Federation established the campaign in 1991.

They picked November 14th because it marks the birthday of Frederick Banting: the individual who discovered insulin.

“Some 422 million individuals across the globe have diabetes.” — World Health Organization

Diabetes is significant to me both as someone dealing with the condition and as a professional, as my team and I continue to develop Suguard: an AI-based smartphone app designed to make daily life easier for individuals with the condition.

Suguard is an internal project we’ve been working on since 2014, the year we established DiabetesLab: our second company focused on creating advanced software that aids individuals in managing an illness using AI.

Suguard is not only my brainchild but also my aspiration. As someone grappling with the condition, I see a substantial need for such a personalized application. My experiences have been the driving force behind my quest to find a solution to help me stay active and enjoy sports.

Individuals with diabetes often require extensive treatment and exceptional care, especially during physical activities. But that does not make it impossible. And I am firm in my belief that I am living proof that individuals with the condition can still engage in sports at a high level .

I am speaking about this because my desire to engage in sports compelled me to create a solution that would assist me. And I hope that soon, it will be the most useful app globally for individuals with Type 1 diabetes.

Should this pique your interest, you can find out more about the project in my article on How AI and Data Science Can Help Manage Diabetes in Everyday Life). However, today my focus is not on Suguard.

Instead, I am here to share other AI-based solutions that are aiding individuals with diabetes in managing the condition.

I hope you appreciate the insights.

Five Methods by Which Artificial Intelligence Enhances Diabetes Care

There are numerous ways to utilize AI for diabetes. The following five are the most innovative applications I am aware of; if you know of any others, please send them my way.

1. Diagnosis of Diabetic Retinopathy

Physicians are effectively utilizing deep learning to automate the diagnosis of diabetic retinopathy: a complication linked to diabetes that can lead to vision loss.

Experts are employing AI-based screening to identify and track occurrences of diabetic retinopathy, with 96% of patients being satisfied with the service. The technology utilizes convolutional neural networks to identify potential issues on a patient’s retina, achieving accuracy levels of 92.3% and specificity levels of 93.7%.

2. Modeling Disease Risk

Healthcare institutions leverage machine learning to create models that predict the likelihood of diabetes within specific population groups. This involves analyzing factors such as lifestyle, physical and mental well-being, and social media activity.

A dataset of 68,994 individuals was utilized to train the algorithm for predicting diabetes, resulting in a highly accurate prediction model. The software not only assesses the risk of long-term complications like Diabetic Retinopathy and cardiovascular or renal issues but also considers short-term concerns such as hypoglycemia.

3. Self-Management of Diabetes

Effective self-management plays a pivotal role in diabetes care. AI has empowered patients to take charge of their own health by using personal data to tailor their lifestyle and essentially assume the role of an at-home healthcare provider.

Artificial intelligence allows individuals to make informed decisions regarding dietary choices and physical activity levels. Smartphone applications like Suguard simplify self-management through real-time analysis of food’s calorific value.

4. Advanced Genomic Studies

Genetic makeup holds valuable insights into one’s health. Advanced molecular phenotyping, epigenetic changes, and the rise of digital biomarkers are aiding medical professionals in enhancing the diagnosis and management of conditions such as diabetes by harnessing genomics.

Microbiome data has provided a wealth of microbial marker genes that can predict the likelihood of diabetes and even guide treatment. Furthermore, research has uncovered over 400 genetic signals that indicate the risk of developing diabetes.

5. Monitoring Complications

Diabetes can lead to various common complications, including vascular disorders (manifesting as strokes, blood clots, or arterial disease) and peripheral neuropathies (resulting in weakness, numbness, and pain, particularly in the hands and feet).

Similar to the use of machine learning in Diabetic Retinopathy diagnosis, AI can aid in identifying and monitoring other related issues. For instance, an app named FootSnap is capable of detecting inflammation and predicting potential foot ulcers.

AI’s Impact on Lives

Artificial intelligence has brought about a significant transformation in the daily lives of individuals affected by diabetes. Abundant disease-related data is not only enhancing self-management but also customizing treatment plans, with a growing number of advanced solutions entering the field each year.

How will AI transform medical diagnostics in 2024?

The healthcare sector will undergo a revolution with the introduction of AI in medical diagnostics in 2024.

Advanced machine learning algorithms will be swiftly integrated into healthcare systems, enabling medical professionals to analyze globally large volumes of patient data to identify patterns that will not only enhance the accuracy of their diagnoses but also help them discover broader and potentially previously unknown connections, leading to earlier detection.

The end result will be improved patient outcomes, reduced workload for healthcare workers, and potentially the identification of new diagnostic techniques.

Here are a few ways we can anticipate the integration of AI into the diagnostic process in 2024:

1. AI-generated and self-diagnosis

Self-diagnosis refers to individuals attempting to diagnose their own illnesses based on their symptoms, typically by consulting online resources. Search engines and social media have historically played a significant role in self-diagnosis, and up to one-third of people in the United States have used the internet to diagnose their own ailments.

Self-diagnosis can benefit the healthcare sector – if patients can accurately diagnose their symptoms, it can alleviate the burden on general practitioners and lead to quicker, better outcomes.

However, one of the major drawbacks of using the internet for self-diagnosis is that patients often misdiagnose their illness, either by misunderstanding the link between a symptom and the associated disease, overemphasizing the significance of one symptom, or overlooking a symptom altogether.

Confirmation bias also plays a significant role: if a patient is convinced they have a specific illness, they may be inclined to omit or fabricate symptoms to align with the diagnostic criteria. As a result, approximately 34% of all self-diagnoses are incorrect, which can lead to complications later on.

This is where Artificial Intelligence comes in. New AI chatbots will have access to an extensive collection of medical literature as well as the ability to develop comprehensive understandings of symptoms and rapidly process data to generate potential diagnoses. This will enable patients to describe their symptoms and receive immediate feedback, aiding them in self-diagnosing with more accurate results.

2. Utilizing big data for predictive analytics

The healthcare sector already accounts for over one-third (33%) of all data worldwide. This data is growing at an exponential rate, faster than in any other sector. In fact, a single hospital in the USA generates approximately 137 terabytes of new data per day. Given the vastness of this pool, it would be practically impossible for human knowledge workers to derive meaningful insights from it.

Fortunately, AI enables the automated handling of healthcare data, including processing and reporting. Through supervised learning and the creation of deep neural networks, healthcare professionals are training AI to understand and interpret healthcare data in order to enhance diagnostics. This involves analyzing extensive data sets , identifying trends within the data, comparing data with other population-wide and historical data sets, and cross-referencing results with decades’ worth of medical literature. Processes that would have taken human experts weeks or even months to complete can now be accomplished by AI in minutes.

At the beginning of 2024, AI is already being utilized in various diagnostic methods, not just for processing textual and numerical data, but also in medical imaging research (such as X-rays, CT scans, and MRIs). For example, by examining the buildup of plaque in a patient’s arteries across sets of computed tomography angiography (CTA) images, researchers at Cedars Sinai have developed an AI model capable of identifying patients at risk of heart attacks.

In addition, researchers are exploring the use of AI in big data analysis to create diagnostic models for conditions like breast cancer, dementia, diabetes, and kidney disease. The goal is for these AI models to automatically identify patients’ risks of various illnesses and initiate treatment before these conditions become critical. In addition to potential cost savings, these preventive treatments could potentially save millions of lives each year.

3. Remote patient monitoring

Another area where AI is impacting the diagnostic process is remote patient monitoring. Currently, triage heavily relies on patients presenting themselves to a healthcare professional while displaying symptoms. This can lead to errors, such as when the symptoms presented at the time do not align with the diagnosis, when the patient is asymptomatic, when the severity of symptoms is misinterpreted, resulting in a more urgent or less urgent response than necessary, or when a diagnosis is missed entirely.

These errors and misdiagnoses can, in turn, lead to wasted time, effort and money. Misdiagnoses are believed to cost the US healthcare industry around US$100 billion per year.

One part of the solution may lie in AI-powered remote patient monitoring, allowing patients to be monitored over time in order to keep track of changes in their health. Remote patient monitoring could pave the way towards more accurate diagnoses by tracking the development, changes , and severity of symptoms over a sustained period of time using a variety of AI-augmented tools, including wearable devices, sensors, and patient-reported information.

Not only could this system be used to catch symptoms that may otherwise be missed, it offers the potential for doctors to spot symptoms earlier, leading to faster diagnoses and potentially better patient outcomes. Better still, in the search for one diagnosis, medical professionals may be able to spot other diagnoses, saving the patient from having to attend triage multiple times.

4. New diagnostic research

Artificial intelligence can now enable healthcare practitioner to identify new diagnostic models. This could apply both to never-before-identified illnesses or variations of existing illnesses, and to new diagnostic frameworks for well-known illnesses.

AI’s ability to process huge segments of data will allow medical experts to spot new patterns and trends developing across a population. This could lead to many interesting benefits. For instance, with virulent diseases, AI will be able to track the spread of these diseases and allow experts to identify how the illness moves from person to person, how quickly it can spread, time to incubation and appearance of first symptoms, and so on.

This methodology was effectively used during the recent COVID-19 pandemic. AI helped to model disease clusters, predicting the likely spread of the illness throughout a given population, and thus informed healthcare experts as to what would be the best possible response.

This led to the development of AI-influenced contact tracing (identifying likely exposures), monitoring and early diagnosis (the ability to work backwards to identify first symptoms), and telemedicine responses (used to inform the likelihood of probable diagnosis without needing to refer individual patients to a healthcare practitioner, thus reducing workload and burden).

Artificial intelligence will bring new, streamlined ways of working to the practice of medical diagnostics.

As we’ve seen, AI has the potential to:
- Speed up the diagnostic process, relieving the pressure on the medical professionals involved in triage
  Allow for earlier diagnosis, both by identifying symptoms that may otherwise go unnoticed, and through patient monitoring, which enables illnesses to be identified even before a patient presents at triage
- Improve the accuracy of diagnoses, by comparing symptoms against a vast compendium of medical literature and big data gathered from other sources to provide suggestions that can be confirmed by a professional
- Model trends across a population by analyzing large data sets and identifying patterns
- Reduce the burden on healthcare workers, leading to cost savings and freeing up experts’ time and resources for more urgent cases
- AI will have a profound impact on the healthcare sector, helping to improve both the efficiency and the quality of medical diagnostics and hopefully producing better outcomes for patients.
However, the rapid development of AI and its integration into the healthcare sector is not without its challenges, some of which include:

Potential for large-scale inaccuracies

Artificial intelligence is a learning model, and much of this learning comes from human-generated data. Indeed, AI itself is programmed by humans. This brings about the risk of inaccuracies, both in the fundamental make-up of AI, and in its ability to process data. AI is also unable to discriminate between good data and bad data, running the risk that even a minor inaccuracy could have massive consequences if AI takes it as fact.

In terms of diagnostics, AI could return large-scale misdiagnoses, prescribe treatments incorrect, or process its own learnings incorrectly. Given the scale that AI works at, the cost of a single bad decision could have far-reaching consequences if left unchecked.

Ethical considerations

As AI becomes ever more integrated into our healthcare system, humanity must reckon with the ethical consequences this may have. For one thing, it is already well-documented that AI exhibits signs of racial and gender bias. But perhaps even more concerning is the fact that artificial intelligence is not capable of human empathy.

This could significantly impact diagnostics, as AI may comprehend a diagnosis medically but not grasp its psychological and emotional effects on the patient. We need to be cautious not to delegate too much of the diagnostic process to AI, risking the neglect of the vital patient- doctor relationship.

Adjusting to global changes

It’s important to recognize that the integration of AI in medical diagnostics signifies a fundamental change for the worldwide healthcare sector. There is a need for extensive preparation, including training, public awareness initiatives, and open communication between medical professionals and patients, to facilitate this major transition.

The effectiveness of AI integration should not be gauged solely by its ability to save time and reduce costs, but rather by its societal impact, the value it adds for individuals, and its level of societal acceptance.

Patients with type 1 diabetes who are receiving insulin treatment and may experience hypoglycemia must notify the National Driver Licence Service (NDLS) and follow the precautions outlined in the Medical Fitness to Drive Guidelines from April 2017. The purpose of this study was to evaluate both awareness and compliance with these guidelines, identify if certain demographics exhibit higher adherence rates, and determine if patients receive counseling from their general practitioners concerning safe driving practices.

In Ireland, the health of drivers is monitored through both European Union laws and regulations established by the Road Traffic Acts in Ireland. The Medical Fitness to Drive Guidelines represent an interpretation of these laws and have been developed based on current medical evidence and established international practices. They outline driving restrictions for various medical conditions, including insulin-treated diabetes.

In Ireland, individuals with type 1 diabetes make up 10-15% of the entire population of diabetes patients, totaling just over 207,000. There is ongoing debate over whether individuals with diabetes experience higher rates of accidents compared to the general public. Existing studies often do not differentiate between diabetes types and rely on patient recall, which indicates a need for high-quality, extensive prospective studies.

Prior research has indicated that healthcare professionals frequently provide insufficient guidance to patients with type 1 diabetes regarding safe driving. While there have been studies published internationally on this subject, significant data specifically from Ireland is lacking.

The primary safety issue for individuals with type 1 diabetes related to driving is hypoglycemia. Increased driving risks are associated with those who frequently endure severe hypoglycemic episodes, those who have previously experienced a hypoglycemic episode while driving, and those who do not check their blood glucose levels before getting behind the wheel.

It seems that patients often decide whether to drive based on their awareness of hypoglycemic symptoms. However, research has shown that relying on symptom-based estimates of blood glucose levels is neither accurate nor safe.

There are evident gaps in knowledge among both patients and healthcare professionals regarding the safe driving recommendations for individuals with type 1 diabetes. Enhanced access to information about reducing driving risks associated with diabetes is necessary for patients who use insulin to become more knowledgeable about driving regulations and recommendations.

Methods

A total of 107 participants were involved in our study, comprising 55 males and 52 females. The participants’ occupations included manual (6), professional (48), skilled workers (20), as well as unemployed (25) and retired (8) individuals. On average, patients in the study had been diagnosed with type 1 diabetes for 18.5 years.

We performed a cross-sectional, quantitative survey using a SurveyMonkey link to a self-created questionnaire. The questionnaires were distributed through diabetes clinics at CUH, GP surgeries, and online diabetes support groups.

Data was recorded in Microsoft Excel and analyzed using SPSS software. The chi-squared test was employed to determine P values for the strength of the association between different study variables. The Clinical Research Ethics Committee of the Cork Teaching Hospitals granted approval for the study.

Severe hypoglycemia while driving

In terms of severe hypoglycemia experienced during driving—defined as an episode requiring assistance from another person—one participant reported having a severe hypoglycemic episode while driving within the past year, and two participants reported two such episodes. One patient mentioned that a previous hypoglycemic episode while driving had led to an accident.

When suspecting hypoglycemia while driving, 11 (10.3%) participants planned to continue driving with heightened caution, 67 (62.6%) would stop driving, remove the keys from the ignition, relocate to the passenger seat, consume a carbohydrate source, and then resume driving, while 29 (27.1%) indicated that they would stop driving, take the keys out of the ignition, move to the passenger seat, eat/drink a carbohydrate source, and rest for at least 45 minutes before driving again.

Discussion

Most participants in this study were conscious of the fact that driving when blood glucose levels are below 5mmol/l is unsafe. This awareness is crucial, as cognitive impairment has been shown to occur when blood glucose drops below this threshold. However, the blood glucose testing practices among drivers with type 1 diabetes are largely inadequate, and a significant number of participants were not compliant with the established guidelines.

It is concerning that 8.4% of patients never keep their testing kit in their vehicle while driving, and only 34.6% consistently check their blood glucose before driving. Fourteen percent do not test their blood glucose before driving, and among the 36 individuals who seemed to understand the guidelines from the Licensing Authority, only 15 (41.7%) said they always monitor their blood glucose level before driving.

It is worth noting that there are no stipulations regarding regular blood glucose monitoring for obtaining a standard driving licence. Nevertheless, neglecting to check blood glucose levels may lead to legal repercussions, as earlier research has indicated, so effective education from healthcare professionals is crucial.

Only 29 (27.1%) of the participants understood the suitable management of hypoglycaemia while driving, indicating that a small percentage of patients in the study were informed.

The study focused on patients with type 1 diabetes for several reasons. From reviewing the literature related to diabetes and driving, it is evident that type one and type two diabetes patients largely represent distinct groups. For instance, individuals with type two diabetes are typically older and often have multiple comorbidities or significant complications from diabetes, such as retinopathy or neuropathy, which can also affect their driving safety.

However, a future study could include patients with type two diabetes undergoing insulin treatment or oral medications that carry a risk for hypoglycaemia, which would likely produce intriguing findings.

Strengths and limitations

A notable strength of the study is that the sample size is comparable to or even larger than other relevant studies in the field, most of which were published abroad, apart from a clinical audit conducted in Sligo Regional Hospital in 2013.

One limitation of the study is that participants were not asked if they had notified the NDLS about their diabetes, raising questions about adherence to legal requirements. Since the data was self-reported, there may have been some bias. Additionally, this study included responses from individuals who actively partake in diabetes support groups, which may suggest that these patients possess greater knowledge about driving regulations compared to individuals with type one diabetes in the broader population.

Conclusion

The risk of hypoglycaemia is a significant concern for individuals with type 1 diabetes. It is essential for health professionals to thoroughly review current driving practices and maximize opportunities to provide information and reinforce safety measures for patients, as outlined in the Medical Fitness to Drive Guidelines.

The clinical importance of this study is to enhance patient care through adequate education and to contribute to the safety of all drivers on the roads.

General practitioners often see patients with diabetes more regularly than other healthcare professionals who are involved in this care area. They must be well-versed in current driving guidelines and regulations for individuals with type 1 diabetes to provide the most accurate and updated information.

The ADA has cautioned against across-the-board driving restrictions, advocating instead for assessments on an individual basis.

The American Diabetes Association (ADA) asserts that having diabetes should not prevent someone from driving, emphasizing that only a medical professional should determine if complications are severe enough to restrict an individual from driving.

A new position statement published in the January issue of Diabetes Care advises against universal bans or restrictions. It suggests that patients facing potential driving risks due to their conditions be evaluated by their regular physician who treats individuals with diabetes.

“There have been inappropriate pressures to limit driving licenses for those with diabetes, and we were worried these recommendations were coming from individuals lacking sufficient knowledge about diabetes and were needlessly overly restrictive,” explained Dr. Daniel Lorber, chair of the writing group that created the position statement and director of endocrinology at New York Hospital Queens in New York City.

“The vast majority of individuals with diabetes drive safely,” noted Lorber. Currently, states have varying laws regarding driving and diabetes, and the ADA advocates for a standardized questionnaire to evaluate driving safety.

Nearly 19 million individuals in the United States have been diagnosed with diabetes, a condition that affects blood sugar levels. The primary concern regarding drivers with diabetes arises from the risk of low blood sugar (hypoglycemia), which may lead to confusion and disorientation. Although a hypoglycemic episode can impair driving ability, the ADA states that such occurrences are uncommon.

An analysis of 15 previous studies on the relationship between diabetes and driving revealed that, in general, people with diabetes have between a 12 percent and 19 percent increased likelihood of being involved in a motor vehicle accident compared to the general population of drivers.

However, society often accepts more dangerous driving situations. According to the ADA, a 16-year-old male faces a 42 times greater likelihood of being involved in a car accident compared to a woman aged 35 to 45. Individuals with attention-deficit hyperactivity disorder (ADHD) have an accident risk that is roughly four times higher than that of the general population, and those with sleep apnea are approximately 2.4 times more likely to be involved in a crash.

“The challenge lies in identifying individuals at high risk and creating measures to help them reduce their chances of driving accidents,” noted the ADA committee.

For instance, people with diabetes who use insulin are at high risk for experiencing hypoglycemia. The ADA advises those on insulin to check their blood sugar before operating a vehicle and to retest at regular intervals if their drive lasts longer than one hour.

“Nowadays, patients with type 1 diabetes are just like everyone else. There’s no justification for limiting their ability to drive,” stated Dr. Joel Zonszein, who leads the clinical diabetes center at Montefiore Medical Center in New York City. “Today’s patients are quite knowledgeable and have access to more technology to manage their diabetes and prevent hypoglycemia.”

For individuals at risk of severe hypoglycemia, the ADA recommends against starting a long drive with blood sugar levels that are low-normal (between 70 and 90 milligrams per deciliter) without consuming some carbohydrates to avoid a drop in blood sugar while driving. The ADA also suggests keeping a quick source of carbohydrates (such as fruit juice, hard candy, or dextrose tablets) in the car to swiftly raise blood sugar, along with having an additional snack like cheese crackers available.

Other diabetes-related factors that could impact driving include diabetic eye disease (retinopathy) and nerve damage (peripheral neuropathy). Retinopathy can impair vision, whereas neuropathy may limit the ability to feel the gas and brake pedals. If these complications are severe, driving could become problematic.

The ADA advises individuals with diabetes who might be a risk while driving to seek evaluation from a physician knowledgeable about diabetes. If their condition jeopardizes their ability to drive safely, doctors can inform state licensing agencies. The ADA does not advocate for mandatory physician reporting, as it could discourage individuals with diabetes from discussing these matters with their healthcare providers.

The key takeaway for those with diabetes, according to Lorber, is to “check your sugar before you drive, and do not drive if your levels are below 70 mg/dL.”

Due to the potential for a substantial decrease in glucose levels in the central nervous system (CNS), the functioning of higher brain centers diminishes, leading to a reduction in cerebral energy requirements.

Hypoglycemic conditions can be induced by medications or substances such as insulin, alcohol, or sulfonylureas. Less commonly, they can be caused by salicylates, propanolol, pentamidine, disopyramide, hypoglycin A, or quinine.

Non-drug-related hypoglycemia can arise from fasting, exercise, tumors, liver disease, severe nephropathy, or have an autoimmune basis.

Symptoms and signs can be adrenergic, presenting as sweating, anxiety, general tremors, palpitations, lightheadedness, and sometimes hunger.

Manifestations affecting the CNS may include confusion, inappropriate actions, visual disturbances, stupor, coma, and seizures.

In the early stages of hypoglycemia in drivers, perception, attention, and sensitivity to contrast in visual fields may be compromised. Additionally, cognitive decline is often linked with visual impairment.

Other symptoms that hinder driving include issues with directional control, lack of focus, drowsiness, fatigue, and prolonged reaction times.

When a diabetic driver begins to experience hypoglycemia symptoms, they have already suffered from impaired driving capabilities, posing an accident risk in certain traffic situations.

Many drivers experiencing hypoglycemia believe they are capable of driving safely; however, upon observation, they often exhibit poor judgment or extremely slow reactions.

Only when a driver with hypoglycemia experiences symptoms like tremors, lack of coordination, and visual disturbances do they decide to halt driving.

Thus, the primary concern for these drivers is cognitive impairment—usually unrecognized by them—that renders them unfit for driving and compromises overall safety.

If a hypoglycemic episode in an unconscious individual is not treated promptly, it may lead to seizures and a genuine deficit in brain energy, resulting in irreversible neurological damage or death.

Guidance for Managing Hypoglycemia
- The indications of hypoglycemia are more common in diabetic individuals while driving compared to other daily activities, which negatively impacts their ability to respond to unexpected situations on the road.
- Drivers with diabetes should be educated to recognize their hypoglycemia symptoms early and know the appropriate actions to take in each situation. Delaying response can increase the likelihood of accidents.
- Acute adrenergic symptoms typically lessen with the consumption of glucose or sucrose.
- When individuals on insulin suddenly experience confusion or act inappropriately, they are advised to consume a glass of juice or water mixed with three teaspoons of sugar.
- It is recommended that drivers keep sweets, candies, sugar cubes, or glucose tablets readily available in the vehicle.
- Most hypoglycemic episodes can be addressed through food containing glucose or sucrose for several hours.
- Nevertheless, for patients on sulfonylureas, hypoglycemia may recur for several days, so these individuals should be advised that even if symptoms improve after consuming glucose or sucrose, they must see a doctor immediately and refrain from driving.
- A hypoglycemic driver who continues to experience confusion and visual disturbances despite taking sugar should not drive and should seek assistance for urgent transport.
- A patient who exhibits CNS symptoms due to hypoglycemia and does not respond adequately to oral sugar needs to be taken to an emergency department for treatment.
- The indications of acute hypoglycemia combined with loss of consciousness prevent an individual from being fit to drive.
- A diabetic individual should not drive if their blood glucose levels drop to dangerously low levels. The doctor will inform them of the recommended blood glucose thresholds pertinent to their individual case.
- The diabetic driver should understand that if they notice a decline in focus, they should immediately pull over and consume carbohydrates.
- They may resume driving only once they fully recover, always ensuring to check 1-2 hours later that their blood glucose levels have not decreased again to unsafe levels.
- Moreover, the recovery time from hypoglycemia to being able to drive safely will vary depending on the trip type, road conditions, and whether they have company in the vehicle.
- Before embarking on a journey, the patient should always check their blood glucose levels, making sure they are within the normal range established by their doctor.
- During trips, meal schedules and medication regimens should be adhered to. It is advisable for the driver to keep sweets, sugar cubes, or glucose tablets in the car.
- Throughout journeys, they should be accompanied by individuals who are familiar with their condition and can provide assistance in case of complications. They should take breaks every hour.
- The driver should keep a visible medical report inside the car that specifies their condition and treatment so that it can be identified and appropriately managed in the event of an accident.
- Drivers should refrain from consuming alcohol prior to driving. Diabetic drivers, in particular, are advised against drinking alcohol due to its potential interference with their medication, thereby increasing risks associated with driving.
November 6, 2024
How to Use ChatGPT for Travel Planning
Artificial intelligence can also be used to plan travel routes and bundle tips for tourists. The industry is following the trend closely. But how does it work in practice? A city guide tried it out.

Brent Foster is curious. The Californian has been working as a city guide in his adopted home of Hamburg since 2010 – he knows the city inside and out. But in the age of ChatGPT, his job could soon be under threat.Tourists who come to Hamburg can also use the artificial intelligence ChatGPT to generate walking routes or put together a table with travel tips.

From a three-week tour of Thailand to a short walk through Hamburg: ChatGPT seems to know its stuff. A threat to tourist experts like Foster? At Hamburg’s Rathausmarkt, the city guide tests the program – with the following “prompt” (ie the instruction): “Tell me a walking tour of Hamburg that takes one hour.”

The request is still quite general, and the answers are just as general – start at Rathausmarkt, continue to Jungfernstieg, along the InnerAlster, return via Mönckebergstrasse and then to the main station to admire its”impressive architecture”. Foster thinks that you can do that, but it’s pretty standard. Is there anything more that can be done?

Statues added at the town hall

Travel planning with ChatGPT is still an insider tip, but is already being used by influencers and travel bloggers. Influencer Diana zurLöwen, for example, recently used the tool to plan a trip to London. Her tip:define specifically what interests you have and what are no-go criteria on a trip.

The more ChatGPT knows about your profile and travel wishes, the better it can respond. It doesn’t have to be a question: “You can also ask counter-questions on ChatGPT, so that you can really have a whole conversation,” says zur Löwen. “It’s really worth trying it out bit by bit.”

City guide Foster is testing such a conversation with the AI at Hamburg City Hall and wants to know from ChatGPT what the statues on the outside facade mean. The answer comes promptly, but is disappointing: of the five figures mentioned, the artificial intelligence has only correctly identified one, but also adds new figures.

For example, the long-time city guide has not yet been able to spot Karl Marx on the facade, and an inquiry to the tourist office also shows that no one here has ever heard of Karl Marx. The digital travel companion is still prone to errors. Influencer zur Löwen advises checking the tips “or just using them as a kind of basis and then thinking about it a bit yourself and checking again.”

TUI plans to use the technology soon

Despite the susceptibility to errors of artificial intelligence, the travel industry is closely monitoring developments and is already developing its own ideas on how the technology could be used. “I see that test projects are being called for everywhere, that there is great curiosity everywhere,” says the chairman of the German Travel Association’s Committee for Digitalization, Oliver Rengelshausen. The topic is being discussed at conferences, association members are being trained, and ideas are being debated.

Some ideas are soon to be implemented at the tourism group TUI. Christian Rapp is the TUI Group’s press spokesman for technology issues and reports, among other things, on AI projects for travel agencies: “In the Netherlands, we are looking at how we can help travel advisors in travel agencies find information more quickly within our own internal information systems.”

The aim is not to replace workers in travel agencies, but AI can help them access information more quickly. The expectation is “that certain tasks will become easier and can be completed more quickly, so that our colleagues in travel agencies actually have more time for what their actual job is: providing personal advice to customers.”

Elbphilharmonie as an “insider tip”

City guide Foster is not worried that he could become replaceable with his Hamburg tours – when he returns from the ChatGPT round, he points to a bright yellow umbrella and a group of tourists in front of Hamburg City Hall: a city tour by a colleague from “Robin and the Tourguides”. Well attended. Foster believes that this personal contact remains irreplaceable.

And: Chat GPT has not yet convinced him; important information was missing from the short tour, mistakes crept in when asking questions, and the route was planned in a somewhat impractical way. Perhaps a tool for getting started in a new city? “You might get a first glimpse of a city you don’t know,” he says.

At the very end, it tests again whether a very precise query might produce better results: What insider tip does ChatGPT have for lovers of classical music in Hamburg? The answer is sobering: The Elbphilharmonie is recommended as an insider tip. But then the artificial intelligence also suggests concerts at the Hamburg University of Music, for example, which are actually more of an insider tip. But travelers will probably still have to be a little patient with the artificial intelligence and experiment a lot.

Embracing the future of AI or watching The Terminator with a sense of foreboding, the rapid rise of ChatGPT cannot be ignored. The platform, owned by OpenAI, allows users to converse with an AI-powered chatbot and gained over 100 million users in three months after its launch in late 2022, sparking controversy. (The number of users fell for the first time in June 2023, indicating decreased initial interest.)

ChatGPT’s rise has forced society to face questions about the role of artificial intelligence. Companies like Bing, Expedia, and Matador Network have quickly adopted AI in travel planning tools.

As someone interested in tech, I feel both doubtful and open-minded about AI’s future. As a travel editor, I wondered if ChatGPT could create a comprehensive travel itinerary or something more concerning.

So I had ChatGPT plan a weekend trip to Washington, D.C., a destination I wasn’t familiar with.

I planned to stay at the Waldorf Astoria Washington D.C., near major attractions like the National Mall, the U.S. Capitol, and the White House. (Although the White House, a recognizable U.S. building, was noticeably absent from my itinerary.) My trip was entirely at the mercy of the robot.

Here’s what I learned about using ChatGPT to plan a trip and if I’d use it again as a travel tool.

If planning is your favorite part of travel, using ChatGPT might take away the excitement of the discovery phase. Researching a destination, browsing social media for recommendations, and scouring Google Maps for hidden gems is what excites me about a trip. With ChatGPT, I missed out on this preparation phase and felt disconnected from my itinerary. The anticipation and payoff I usually get from visiting a new place were essentially absent with ChatGPT as the planner.

ChatGPT can help you get organized if used correctly. AI can be a solid planning partner for the travel planning enthusiasts. For example, ChatGPT was the “big picture” guy for major stops while I managed the detailed itinerary during my trip to D.C. In another instance, asking ChatGPT to create a logical route for my trip to Iceland’s Westfjords helped me get organized. In this case, I was the big-picture planner and ChatGPT helped with the details.

Using ChatGPT takes practice. Like any tool, mastering ChatGPT will take time, and crafting a query that covers all bases may take a couple of tries. Your opinion of the tool will depend on your patience level. For some, it may be a fun puzzle to solve, while for others, it may become tedious, especially with the need to fact-check and adjust the schedule. Being specific with your ask will help ChatGPT tailor an itinerary to your needs. Details such as travel dates, interests, accommodation, budget, group size, and if it’s your first time visiting the destination are essential.

Suppose, for instance, here was my final request for planning my trip to Washington, D.C.:

Hello ChatGPT! My partner and I plan to visit Washington, D.C. from July 6 to July 8. Can you put together a 2-day travel plan for us that includes restaurants, bars, and places of interest based on the details below?
- This will be our first time in D.C.
- We’ll be staying at the Waldorf Astoria DC.
- Our arrival is scheduled for 1 p.m. on July 6, and we’ll be leaving at 4 p.m. on July 8.
- We are in our mid-20s and are enthusiastic about art, history, food, and music.
Please be aware that ChatGPT might not always provide accurate information, so additional research is required.

Using ChatGPT as a travel planner has its downsides, mainly due to the possibility of inaccurate information. The latest model, ChatGPT-4, which is available at a cost of $20 per month, was last updated in March 2023, while the free version has not been updated since September 2021. This means that a suggested itinerary may include closed businesses or outdated entrance fees and hours of operation.

It’s also important to note that ChatGPT is not adept at factoring in travel times or creating an efficient timetable unless specifically requested. During this trip, I found myself moving between neighborhoods rather than following a logical itinerary. While travel times of 20-30 minutes on the train here and there may not seem significant, they can quickly accumulate, causing disruptions to your schedule and potentially leading to fatigue.

While ChatGPT can provide decent recommendations, it is essential to verify opening hours, ticket availability, reservations, and potential impacts of factors such as local holidays or temporary closures on your travel plans. (I discovered this the hard way when I arrived at the African American Civil War Museum in D.C.’s sweltering midsummer heat, only to find the indoor exhibition had been closed for renovations since March.)

At the end of each itinerary generated by ChatGPT, there is a reminder that all itineraries should be fact-checked. However, if you miss this warning or choose to trust the AI without reservations, you may end up with an itinerary that overpromises and underdelivers.

ChatGPT ensures that you cover the essentials . . .

One thing that can almost be guaranteed with ChatGPT is that you won’t miss out on the must-see attractions. Except for the White House, my itinerary included the major attractions that any first-time visitor to the nation’s capital would want to visit, such as the Smithsonian Institute, the National Mall, the African American Civil War Memorial, the Library of Congress, and the Capitol Building. In addition to the major tourist attractions, D.C. institutions like Ben’s Chili Bowl and the 9:30 Club, an iconic music venue that has been around for decades, were also included in the list.

While none of these recommendations were surprising, I felt that I was making the most of my relatively limited time in the city. If your goal is to see the highlights, ChatGPT will prioritize getting you there.

. . . but more interesting recommendations and advice are likely to come from a human

Apart from the essential stops, the bars and restaurants suggested by ChatGPT were good, but not exceptional. I did not come away convinced that AI can rival, or even match, recommendations from another human, whether through word of mouth, a travel website, or a Reddit thread on “Best things to do in ____?”

One of my friends, who visits the capital several times a year, mentioned that ChatGPT’s list was fairly good “for people who are only going to go to D.C. once and aren’t looking for any niche experiences” and shared a few suggestions that I found more appealing from the outset.

Another friend, who currently resides in D.C., noted that the itinerary seemed too packed to be enjoyable, and the order of the itinerary “was not ideal in terms of economical travel,” two major points that I also observed.

Overall, seeking recommendations from a person, especially someone you trust to provide solid suggestions, seems to offer a higher likelihood of discovering new openings, local favorites, or hidden gems compared to asking a bot for suggestions.

ChatGPT does not account for the “human element”

It’s rather obvious, but worth stating that ChatGPT is not human and therefore cannot consider the “human element”—those small factors that can derail travel plans. It cannot anticipate how worn out you might be after going from one attraction to another, or the impact of crowds, or sudden changes in weather such as summer heat or rain that could render an itinerary full of outdoor activities impractical. Even if you are initially satisfied with your itinerary, it’s wise to have a backup plan in case the ChatGPT-generated plan goes off track.

My verdict on using ChatGPT for trip planning

As AI travel tools advance, I will continue to test future technology, but at present, I probably wouldn’t use ChatGPT to plan a trip again. Despite lukewarm recommendations and the so-called “human elements,” I found that I invested as much time in crafting a query, fact-checking, and adjusting my schedule as I would have if I had created an itinerary entirely on my own—minus the usual enjoyment of planning a trip by myself.

In the not-so-distant future of AI-powered technology, a vacation might kick off by telling your phone something like: “I want to go to Los Angeles for a four-day trip in June, when airfares and hotel rates are most favorable, utilizing loyalty rewards points. I’d like to visit a history museum and an amusement park and have dinner reservations at 7 p.m. near the hotel at a restaurant offering vegan options and a great wine list.” And voila, your phone generates the perfect itinerary.

However, for now, travelers using ChatGPT—the powerful new A.I. software already dabbling in creative cocktail recipes and crafting college papers—may need to manage their expectations.

Oded Battat, general manager at Traveland, a Bridgeport, Conn. travel agency, tried out ChatGPT to find potential excursions for clients traveling to Tuscany as part of his work. He received a list of 14 activities, from winery tours to museum visits, with a suggestion to enjoy gelato in the town square of the medieval hill town San Gimignano. “I was already familiar with all these things,” Mr. Battat remarked, but ChatGPT spared him the trouble of compiling the information and presented it in a format he could easily email to a client.

ChatGPT, the service Mr. Battat started using, made its debut in November and has already begun to revolutionize tech-driven industries, including travel. Distinct from the A.I. most consumers are accustomed to—think website chatbots—ChatGPT is “generative,” capable of analyzing or summarizing content from an extensive array of information sources, such as web pages, books, and other literature available on the internet, and using that information to create new, original content. Its sophisticated natural language capabilities enable it to understand and respond more conversationally.

Numerous applications, as well as limitations

The travel industry may undergo a significant transformation. Already, travelers can interact with the system, sharing details like their destination, time of year, and interests, and in return receive a personalized itinerary complete with vibrant descriptions.

A recent request from a reporter for a two-day itinerary to Whistler, British Columbia resulted in ideas such as guided snowshoeing to observe local flora and fauna and a dog-sled ride “with a team of beautiful huskies” for a winter trip. Upon adding further preferences, like a craving for Thai food, ChatGPT adapts its suggestions, providing new restaurant recommendations based on these specifications.

However, ChatGPT does have its limitations. Initially, its information database only extends to 2021, and it lacks access to critical, real-time travel-related data, such as airline schedules and weather forecasts. New versions are currently in development, with a major upgrade released recently, and further improvements are expected. Additionally, the software doesn’t always discern between reliable and unreliable internet information, sometimes producing inaccurate responses. OpenAI, the creator of ChatGPT, also warns that the software may occasionally yield “biased content.”

The software is available for anyone to use, accessible for free through the OpenAI website. Tourist bureaus can engage ChatGPT to produce marketing content describing must-see attractions, while travel advisors can utilize it to compose emails to their clients and create social media posts. Airlines, hotels, and rental car companies could integrate it to enhance their virtual agents’ ability to handle a broader range of queries.

One travel advisor mentioned using ChatGPT to craft a “firm but amicable breakup letter” to a client with whom she no longer wished to work. Although the advisor had to refine the prompt (the term for a ChatGPT question or command) a few times to achieve her desired outcome, ultimately, it was successful. “My client said she understood and wasn’t upset with me,” mentioned the advisor, who opted to remain anonymous as she didn’t want her former client to know that ChatGPT had crafted the letter.

A ‘significant new step’

Some individuals in the industry are concerned that advancements in systems like ChatGPT may lead to the displacement of travel advisers, according to Chad Burt, co-president of OutsideAgents, a company based in Jacksonville, Florida, with a network of 8,000 advisers. However, Burt believes that the downfall of travel agents has been anticipated before, and each new technology is simply a tool that can be utilized. He recently conducted a tech tips seminar for his advisers and is in the process of compiling a list of prompts that his advisers can utilize to maximize the software’s potential.

Burt, who has been experimenting with ChatGPT, has used it to generate over 100 itineraries. He noted that it serves as an excellent starting point and can save time on basic tasks, but he emphasized that a competent agent still needs to verify and enhance it. According to Burt, only a human can accurately discern what travelers indicate they desire versus what they genuinely want. The software achieves around 70 or 80 percent accuracy, but Burt stressed that they aim for superior quality.

Expedia, a major online travel company, has been employing A.I. for a number of years to customize recommendations and to power its online virtual adviser. However, ChatGPT represents a “significant new step,” according to Peter Kern, Expedia’s CEO.

Kern sees the new technology as a potential method for offering customers a more conversational way to engage with Expedia, for example, by speaking or typing queries instead of clicking. Expedia also envisions leveraging ChatGPT to refine personalized recommendations by merging its data with customer purchase history, airline tickets, hotel availability, and rental car prices.

Aylin Caliskan, a computer science professor at the University of Washington, who specializes in machine learning and the societal impact of artificial intelligence, predicts that other travel companies will adopt a similar approach, integrating their own data and programming with generative A.I. systems developed by companies like Google, Amazon, and OpenAI to achieve specific objectives.

According to Caliskan, creating these systems entails significant investment, data, and human effort, making it more efficient to build upon them. For instance, a travel insurance company could develop a system using the natural language capabilities of software like ChatGPT to assist travelers in selecting suitable policies or navigating the claims process.

Generative A.I. could also enhance foreign language translation, facilitating conversations with locals, according to Dr. Caliskan. When combined with virtual reality technology, it could enable travel companies to offer customers a virtual “visit” to a destination using a virtual reality headset, all without leaving their homes.

Concerns regarding an ‘A.I. junk land’

Jeff Low, CEO of Stash Hotels Rewards, a company that offers loyalty points for staying at a group of independent hotels, is concerned about the impact of new A.I. like ChatGPT on the lodging industry. If the potential of artificial intelligence includes automating routine tasks to allow staff to personally connect with guests, Low believes the reality may be different. He mentioned that hotels have been inclined to reduce staff when A.I. was introduced, such as cutting front desk personnel with the popularity of automated check-in. He stressed that personal interaction is a crucial aspect of travel, and that hotels can distinguish themselves through these connections.

Low also worries that unscrupulous companies could exploit software like ChatGPT to devalue guest reviews on travel sites, which many rely on for making hotel choices. This type of software could potentially facilitate more sophisticated fake reviews, even creating traveler profiles to produce seemingly legitimate reviews over a period of time. While travel companies have systems to combat fake reviews, Low raised concerns about the difficulty in distinguishing legitimate reviews from automated ones.

As more travel providers leverage the capabilities of generative A.I., there are potential downsides to consider. According to Burt, natural language responses can sound very authoritative, leading people to place more trust in them than they should. Furthermore, due to Google’s preference for fresh content when ranking search results, companies aiming to boost their online presence may turn to ChatGPT-like software to generate a growing number of blog and social media posts. Burt believes that this trend could potentially lead to an “A.I. junk land” on the internet.

Despite potential issues, AI-powered advancements could greatly benefit travelers. Chekitan Dev, a professor at Cornell University’s Nolan School of Hotel Administration, suggests that if systems like ChatGPT have access to real-time data, they could seamlessly adjust plans in response to sudden changes. For example, if your flight is delayed, the system could automatically postpone your car rental and inform a restaurant of the need to reschedule your reservation.

The future might bring an autonomous vehicle that anticipates your delayed arrival at the airport, takes you sightseeing, and ultimately guides you to the best pad Thai in town. Another possibility is that AI and virtual reality experts team up to create an almost lifelike vacation experience akin to the “Star Trek” Holodeck, allowing us to travel without leaving home, which is an unexplored domain, according to Dr. Dev.

Artificial intelligence has made its presence known and is shaping discussions. Tech companies are racing to develop AI technology for widespread use, with companies like OpenAI launching the AI chatbot, ChatGPT, last fall. The travel industry has been abuzz with speculation about how these platforms will impact future travel planning.

While some in the travel industry worry that AI technology could replace travel advisors, others are embracing it as a means to enhance the travel planning process.

Can AI streamline vacation planning, allowing you to do it all on your own in record time? Will ChatGPT be up to the task, or is working with a travel agent a better option?

Let’s examine the advantages and disadvantages of using ChatGPT for travel planning.

Although AI software has been a hot topic of discussion lately, with ChatGPT leading the way, some people may not be familiar with the platform.

ChatGPT is an AI-driven chatbot and natural language processing tool that engages in human-like conversations based on user-submitted prompts.

For example, if you’re planning your first trip to Accra, Ghana and aren’t sure where to start, ChatGPT can offer instant advice on the best places to stay, eat, party, and explore, as well as tips to help you save money, avoid crowds, and maximize your trip.

It’s important to note that while the chatbot is useful for travel, it’s also a versatile tool for various purposes. Professionals are using ChatGPT to generate content, write essays, and create cover letters for job applications.

The Benefits of ChatGPT for Travel

Previously, finding the best travel destinations and activities in a new location involved sifting through reviews and conducting extensive searches on search engines. ChatGPT now makes this once-time-consuming task virtually effortless, saving time and effort.

Access to a Wealth of Information

Chat GPT’s strength lies in its ability to process vast amounts of information and deliver detailed responses.

With just a few keystrokes, you can quickly compile a list of activities or accommodations. Instead of combing through multiple pages on a booking website, you can simply provide ChatGPT with your criteria, and it will promptly respond, most of the time.

Lightning-Fast Responses

ChatGPT’s real-time responsiveness is impressive and quite engaging. Simple queries can be answered in as little as 10-20 seconds, while more specific requests may take a bit longer.

When Travel Noire asked the chatbot to create a four-day itinerary featuring Black-owned businesses, it provided recommendations for Black-owned restaurants, bookstores, and neighborhoods in just 60 seconds.

Therefore, ChatGPT can save hours of scouring the internet for activity ideas. While some people enjoy the planning process, freeing up time in this manner allows for other tasks.

Detailed Responses Simplify Itinerary Planning

In Travel Noire’s experiment, ChatGPT produced a comprehensive schedule for a four-day trip to Los Angeles, tailored to the request for a Black-owned experience. The suggested itinerary includes soul food restaurants, cultural arts centers, and even schedules each day by morning, afternoon, and evening. It not only contains an exciting list of top Black-owned businesses in LA but also provides brief descriptions for each business.

How do I request ChatGPT to create a travel plan?

To get the best outcomes, make sure to be as precise as possible when asking a question. The more details you can provide to ChatGPT about your inquiry, the better the feedback you’ll receive. Also, don’t hesitate to ask intricate questions. The AI is designed to learn from being tested. It also learns from the continuous queries of each user, so asking more questions is beneficial. Examples of excellent questions to ask include:
- What are some excellent culinary tours in (mention a city/country)?
- Craft the optimal travel plan for (mention a place) within a $600 budget.
- What are some essential foreign expressions to learn when visiting (mention a place)?
- How much money should I budget for excursions in (mention a place)?
- Can you create a road trip itinerary from (mention a place) to (mention a place)?
- What are the top historical attractions to visit in (mention a place)?
- What is the most suitable means of transportation and route for (mention a place)?
Where AI Lacks

AI tools like ChatGPT can support in sifting through the vast array of travel recommendations available on the internet; nevertheless, there are a few noteworthy areas where the technology falls short compared to humans — at least for now.

Good for Planning, Not as Effective for Booking

At present, the current edition of the application has constraints in its booking functionality. Chat GPT is expanding the platform to enable users to make travel arrangements through third-party booking platforms, but for the moment, options are limited. For instance, Expedia now offers a plugin that integrates with ChatGPT, allowing users to convert their chat responses into real travel bookings.

In comparison to working with a travel agent, arranging travel plans is more do-it-yourself than hands-off. Currently, travel agents still have an advantage because personalized customer service from a human cannot be replaced. Collaborating with a travel professional can aid in creating a trip tailored specifically to your preferences. Moreover, in the event of an emergency or change of plans, a travel agent can provide guidance on cancellations or rescheduling.

Although the platform excels in planning, the journey toward a fully automated AI travel experience will be lengthy.

Restricted Planning Abilities

Currently, Chat GPT can only facilitate travel planning through its comprehensive recommendations and integrations with third-party booking services. Unlike interacting with a human, customization based on your individual interests might be limited. The intricacies of your and your group’s travel preferences may not be fully captured within the technological framework.

You might inform the chatbot that your uncle has difficulty walking long distances, so you require a centrally located place. While you might receive a reply with suitable suggestions, working with industry professionals is still preferable for a truly personalized itinerary.

Platform Overload Problems

With its current popularity, occasional traffic surges can lead to chatbot unavailability. The high demand and overwhelming interest can intermittently cause network errors on the site. This situation can be frustrating for individuals seeking travel insights when the site is at capacity.

Undoubtedly, the potential for how ChatGPT can enhance your travel planning is limitless. Consider giving it a try the next time you’re responsible for planning your group’s travels.

If you’re looking to embark on a vacation but are unsure where to begin, OpenAI’s ChatGPT can offer more than just a bit of support. ChatGPT has emerged as one of the most widely used language processing models of the decade, and people are discovering increasingly clever applications for it. It can assist in constructing code for a specific task and troubleshooting faulty code, aids in planning daily schedules, and much more.

Another progressively popular use for ChatGPT is in vacation planning. There are several ways the service can be employed to assist in creating a vacation plan, from giving destination recommendations to aiding in crafting a budget. Prior to getting started, an OpenAI account must be created, which is required to utilize the tool. It is available for free, and users seeking additional features can upgrade to a $20/month subscription plan for added benefits.

Naturally, it’s important to note that ChatGPT’s suggestions serve as a starting point, and all plans should be diligently verified.

Ways ChatGPT Can Massively Boost Your Efficiency

Request ChatGPT To Serve as a Travel Consultant

Engaging ChatGPT to function as a travel consultant

Users should initially request ChatGPT to act as a travel advisor. Since the tool can adapt various conversational styles and tones, asking it to converse like a travel consultant establishes a context and yields pertinent responses. After inputting the prompt, “Assist me in planning my next vacation as a travel advisor,” the tool responds, “Certainly! I would be delighted to aid you in planning your upcoming trip.” Subsequently, the language processing tool poses questions to assist users in planning their vacation, starting with their destination preferences, travel dates, duration, interests, activities, budget, and other pertinent details.

Find a Destination Based on Your Preferences

Inquiring ChatGPT to recommend travel destinations for a vacation

The first thing ChatGPT inquires about is destination preferences, such as where the individual would like to go and if they have a specific country or region in mind. Users can describe the characteristics of the desired destination, even if they are unsure of a particular city or country at the moment. For instance, someone in a warm region might want to spend a few days in a place with a pleasant climate, away from the city’s hustle and bustle.

Users can also specify whether they wish to travel abroad or stay within their own country. They can also convey if they prefer a place with scenic views or a wide range of recreational activities, or if they simply want to relax and savor delicious cuisine throughout the day. If users are unsatisfied with the initial list of recommendations, they can request ChatGPT to suggest alternate destinations.

Get Acquainted with a Destination before Departing

Asking ChatGPT to furnish additional details about a specific vacation destination-1

ChatGPT can also articulate why someone should or shouldn’t visit a particular place based on their interests. For instance, Screen Rant requested the tool to provide more details about Interlaken, Switzerland. In its response, ChatGPT elaborated that the region is among the most beautiful and vibrant places to visit. It highlighted that visitors can engage in adventure sports, hiking, water sports, embark on nearby excursions, and admire the natural Alpine beauty in the area.

Choose the Optimal Time for Your Visit

Asking ChatGPT when is the ideal time to visit a specific location

Upon selecting their destination, ChatGPT can also assist in determining the optimal travel dates, especially if the user’s schedule is flexible. The chatbot can suggest the most favorable time to visit a specific location based on various factors such as climatic conditions, tourist influx, and more. In the given example, the chatbot recommended that the “best time to visit Interlaken and enjoy pleasant climatic conditions is during the summer months from June to August.”

Users can also provide the duration of their upcoming vacation, followed by their preferred dates. ChatGPT will inquire about any specific preferences related to activities that the user would like to incorporate into their itinerary, enabling it to tailor its responses accordingly.

Explore Activities in the Vicinity

Asking ChatGPT about activities to engage in during a vacation

Screen Rant inquired ChatGPT to propose enjoyable activities available in or around the region, particularly those that are safe to partake in with a group of four to five individuals. In its response, the language processing model included activities such as paragliding and skydiving, which cater to thrill seekers. It also suggested other activities like hiking, boat cruising, rafting, and biking along scenic routes in and around Interlaken. ChatGPT also advised users to verify the credentials and reviews of the tour operators organizing these activities for a better understanding of the experiences.

Set Your Budget and Plan Accordingly

Inquire with ChatGPT about the potential expenses for the trip

To kick off your expenditure planning for your vacation, ChatGPT will require the “specific budget range for the vacation.” During this stage, users should input their trip budget or request an estimation about the potential cost of a trip to the specified destination for the mentioned number of travelers. The chatbot will consider factors such as lodging, transportation, activities, dining, and personal preferences. It concluded that the trip might range in cost from $2,500 to $6,000, not including international flights.

The language processing tool can provide a rough estimate of travel costs depending on the preferred mode of transportation, whether it’s public transport or a rented cab. However, it’s worth noting that the chatbot’s database might not be up to date, so it’s advisable to further validate its suggestions with additional research.

Seek Accommodation Advice

Ask ChatGPT for recommendations on accommodations for a trip

ChatGPT can also offer users suggestions for well-known lodging options. Upon asking the chatbot, “please recommend some budget-friendly accommodation in Interlaken, Switzerland,” it presented various options. While it correctly identified the names of the places, the price range it provided differed from the actual rates available online. In the response, ChatGPT indicated that “these are estimated costs and can vary based on factors such as the time of year, availability, and specific room types.” Therefore, while users can gather some recommendations, it’s best to verify this information for a more precise understanding.

Create a Travel Plan

Ask ChatGPT to generate an itinerary for a trip

Finally, request ChatGPT to put together an itinerary containing all the details, including the daily schedule, travel times between locations, and other relevant information. Initially, the tool responds with a paragraph, but users can request it to design a table with multiple columns for ease. Although ChatGPT cannot export the itinerary into a spreadsheet, users can always capture screenshots and obtain a hard copy or save a digital version on their mobile device or tablet.

In this specific example, the itinerary included all the necessary information, but again, some details appeared slightly inaccurate, so users should always double-check. For instance, the language model states that a flight from Lucknow, Uttar Pradesh, India, to Zurich , Switzerland, takes 10 to 12 hours, while the current fastest flight actually takes more than 13 hours. However, the approximate travel time from Zurich to Interlaken is correct.

While it’s not advisable to rely solely on ChatGPT, its suggestions can serve as a helpful starting point for planning a vacation.

The era of navigating numerous websites to arrange travel or plan vacations may soon be outdated due to the increasing use of artificial intelligence tools like ChatGPT.

AI can swiftly analyze and gather information from various sources online, delivering responses that resemble those of a human, thereby offering a comprehensive solution for travelers looking to determine attractions, accommodations, and dining options for their journeys, according to Jangwoo Jo, Ph.D., an assistant professor at Metropolitan State University of Denver’s School of Hospitality.

When ChatGPT-4 was launched last year, itinerary planning was highlighted as one of its key features, Jo noted, identifying himself as someone who readily embraces new technology.

“This tool is extremely beneficial for the hospitality and tourism sectors,” he mentioned. “I believe this technology will become widely utilized in our everyday lives, especially for trip or travel planning.”

A significant attribute of large-language models, such as ChatGPT, is their contextual awareness, which means they can grasp the essence of what is being requested, Jo explained. “It comprehends the context: ‘I find myself in this situation. What is a possible solution? What do I need to know?’ This ability enables it to provide tailored travel information.”

“Thanks to context awareness, personalized suggestions that enhance a customer’s experience in the hospitality industry can be optimized,” Jo stated.

To illustrate the capabilities of AI-assisted travel planning in a recent Zoom interview, Jo opened ChatGPT-4o, the latest version of the platform, and posed various questions regarding a hypothetical monthlong trip to his hometown, Seoul, South Korea. The platform generated recommendations covering nearly all aspects of travel.

Flights and attractions

ChatGPT presented a selection of flights from Denver to Seoul found on Expedia, Kayak, and Momondo, and subsequently offered potential lodging options for a monthlong stay. When Jo inquired, “What are the must-see attractions and activities in Seoul during July?” ChatGPT promptly suggested several local historical sites, including the Namsan Tower.

Food and transport

Jo also requested recommendations for restaurants and places to buy cooking supplies, and the platform provided insights on navigating Seoul’s public transportation system.

Jo concluded that ChatGPT-4o was largely accurate. “It has a broad general knowledge of tourist spots,” he remarked.

Language

Finally, he asked, “What are some essential Korean phrases and cultural tips for visiting Seoul?” ChatGPT-4o provided a list of greetings and basic phrases, such as the Korean words for “hello,” “goodbye,” “thank you,” “please,” and “Do you speak English?”

Generative AI models can “understand” as many as 100 languages, enabling them to analyze customer reviews and other written content in those languages, Jo noted.

Booking

Jo did have one exception: “While most of the data is generally reliable, it does not offer a specific option to actually finalize the plans, so you still have to participate in making reservations, processing payments, negotiating prices, and organizing the trip,” he stated.

This could change in future versions of ChatGPT, he suggested, which could pose challenges for online travel platforms like Expedia and Kayak as the AI learns to handle bookings on behalf of users.

“I believe that in the future, generative AI tools will be able to make those reservations and transactions autonomously,” Jo stated. “These online travel agencies are in significant jeopardy. They need to quickly incorporate this AI capability into their systems before AI tools fully integrate online travel services within them.”

When Jason Brown planned his summer vacation to Amsterdam and Ireland this year, he opted not to consult travel books or browse Instagram.

Instead, the founder of recruitment firm People Movers turned to ChatGPT, OpenAI’s generative artificial intelligence tool.

He asked the AI numerous questions to assist in crafting an itinerary for his 10-day trip to Amsterdam and Ireland, covering Dublin and Galway, which he took in July and August with his wife, their two sons aged 20 and 16, and one of their friends.

“In the past, I would always rely on websites like TripAdvisor, but I realized that I had all the information at my disposal [through AI], and it provides results in 15 seconds.” He described the experience as “fantastic.”

“It produced a golf itinerary for Dublin and a four-day plan for the rest of Ireland. It was incredible how it broke it down into morning, afternoon, and evening activities.

“For instance, on the first day, it recommended arriving in the morning, visiting Trinity College and Grafton Street in the afternoon, and then going to Temple Bar in the evening.” Regarding Amsterdam, he noted that it listed key attractions such as the Anne Frank Museum, the Van Gogh Museum, and the Jordaan district. As his trip plans evolved, he continued to refine his queries on ChatGPT.

While he took up many of the AI suggestions, Mr Brown says he still relied on world of mouth recommendations through an online community of people who attended the same college as his, while a friend they visited in Amsterdam showed them around.

“That way we experienced a few things we wouldn’t have found using ChatGPT. But it gives a perfect skeleton of a trip, and gives you everything you need and want to see.”

AI is pervading all areas of our life and travel is no different. As well as ChatGPT there are other generative AI tools such as Google’s Gemini, Microsoft’s Copilot, and dedicated travel AI sites such as Trip Planner and Ask Layla.

It appears to be becoming part of the travel organisation plans for some, with one in 10 Britons having used AI for travel planning, according to a survey by Sainsbury’s Bank Travel Money. One in five said they are likely to use it in the future.

However, the study also suggested that travel AI still has some way to go before it can take on all your holiday plans.

It found that of those who had used AI for travel planning, more than a third (38%) said that it brought up generic answers, 37% said it had missing information, while 30% said it had incorrect information.

While generative AI can help deliver personalised travel itineraries and recommendations, it is only as good as the information it is trained on, and where this information is out of date, biased, erroneous, false and so on, then the AI will perpetuate the misinformation, points out Caroline Bremmer, head of travel and tourism research at analysts Euromonitor International.

“The challenge is ensuring real-time information that is factually correct. There are dangers if consumers do not undertake due diligence to verify the results provided by Gen AI with other sources, including talking to people in the know, such as local residents or travel agents.”

Sardar Bali is the co-founder at Berlin-based AI travel planner and guide Just Ask Layla.

He says accuracy is a key part the service.

“We have internal tools,” says Bali. “All content goes through a two-step verification process, one of which is more automated, and we have a more manual process where internal teams look at different content and researches it a bit.”

But he admits some content “might slip through”.

“For example, it once mentioned an Eiffel Tower in Beijing; it might be tagged incorrectly. But it’s getting better and better every day.”

That improvement is likely to come, particularly as more services come online.

Earlier this year, travel giant Expedia launched an AI service for US customers. Called Romie, it’s part of the company’s iPhone app.

“A trip can involve complex planning… there’s gazillions of options,” says Shiyi Pickrell, senior vice president of data and AI at Expedia Group.

She says Romie can help narrow down the choice of destination, and compare different locations. If you want a beach theme, it can compare British beach destinations to Spain and France for example, or look at which ones are family-friendly.

However, AI doesn’t always go to plan.

Rebecca Crowe, 29, a freelance writer living in Liverpool, says she often taps into AI to help plan her trips, but proceeds with caution after several unhelpful experiences including a trip to Lecco, a town located next to Lake Como in Italy.

“The experience wasn’t great,” says Crowe. “It listed all the popular stuff to do that you’d find with a standard Google search, and the itineraries didn’t make a lot of logical sense.

“They tried to have us in Milan in the morning and Bellagio in the afternoon, and with the train timetables and ferry schedules, this would not really be feasible. It then had us back in Milan the following day to explore more. Following this itinerary, we’d have spent more time on transport than anything else.”

She’s also referred to AI to find gluten-free restaurants when travelling with a friend who has coeliac disease.

“This pulled back results that were massively out of date and just wrong in some cases. I found myself having to manually cross-reference each suggestion to see if the place was even still open.

“If I’m looking for seasonal things like ferry timetables in the shoulder season [months around the peak season], AI just doesn’t seem to be up-to-date and accurate enough. Same for museums with seasonal opening times.”

Instead she advises people to only use it as a sounding board for broad inspiration. “You can find blogs and websites with complete guides and itineraries that are a lot more reliable and up-to-date. If you want a rough idea of things to do in a certain city, it’s a great jumping-off point, but the amount of fact-checking it requires means that it doesn’t really save you much time in the long run.”

Organizing a getaway should ideally be enjoyable. However, compiling a list of activities for a journey can also prove to be time-consuming and stressful, especially if you’re uncertain about where to start.

Fortunately, technology companies have been vying to develop tools that assist with that. Travel has emerged as one of the most favored applications for AI, which Google, Microsoft, and OpenAI prominently highlight in their demonstrations, while companies like Tripadvisor, Expedia, and Booking.com have begun to introduce AI-driven vacation-planning solutions as well. Although fully automated AI agents that can oversee every aspect of planning and booking your vacation are still not quite here, the current generation of AI tools is still quite effective at assisting with various tasks, such as creating itineraries or enhancing your language abilities.

AI models can sometimes generate inaccurate information, so it’s essential to verify their recommendations yourself. Nonetheless, they can still serve as a valuable resource. Continue reading for some suggestions on how AI tools can simplify your planning process, giving you more leisure time to enjoy your trip.

Determine possible destinations for your getaway

First and foremost: you must decide where to go. The advantage of large language models (LLMs) like ChatGPT is that they are trained on extensive amounts of internet data, allowing them to process information that would take humans hours to research and quickly summarize it into straightforward paragraphs.

This makes them excellent resources for generating a list of potential places you might want to visit. The more detailed you are in your request, the better—for instance, informing the chatbot that you’re looking for recommendations for destinations with warm weather, family-friendly beaches, and vibrant nightlife (like Mexico, Thailand, Ibiza, and Australia) will yield more applicable options than ambiguous requests.

However, given AI models’ tendency to produce incorrect information—referred to as hallucinating—it’s advisable to verify that their details about suggested locations and potential activities are indeed correct.

How to utilize it: Activate your preferred LLM—ChatGPT, Gemini, or Copilot are a few available models—and request it to recommend travel destinations. Include key information such as desired temperatures, locations, duration of stay, and activities of interest. An example would be: “Provide a list of destinations for two travelers embarking on a two-week vacation. The locations should be warm during July and August, situated in a city but easily accessible to a beach.”

Select attractions to explore while you’re there

Once you’re on holiday, you can use platforms like ChatGPT or Google’s Gemini to create day trip itineraries. For instance, you might use a request such as “Create an itinerary for a day of driving through the countryside around Florence in Chianti. Include several medieval villages and a winery, and conclude with dinner at a restaurant that has a nice view.” As with LLMs, being as detailed as possible enhances outcomes. To be cautious, it’s wise to cross-check the final itinerary with Google Maps to ensure that the suggested order is logical.

In addition to LLMs, there are also specialized tools that can assist you in assessing the types of conditions you may face, including weather and traffic. If you’re planning an urban getaway, you might want to explore Immersive View, a feature Google Maps introduced last year. It employs AI and computer vision to create a 3D representation showing how a specific spot in a supported city will look at a particular time of day up to four days in advance. By leveraging weather forecasts and traffic information, it can help you determine whether a rooftop bar will be sunny tomorrow evening or if choosing an alternate route for a weekend drive would be wiser.

How to utilize it: Verify if your city is included in this list. Then, open Google Maps, navigate to your area of interest, and select Immersive View. You’ll see an interactive map with options to adjust the date and time of day you wish to examine.

Checking flights and lodging

After deciding on your destination, the next step is to book your flights and accommodations. Many travel booking platforms have incorporated AI chatbots into their services, most of which utilize ChatGPT technology. However, unless you’re particularly loyal to a specific site, it might be beneficial to consider a broader perspective.

Searching for flights across multiple browser tabs can be tedious, but Google’s Gemini offers a solution. This model connects with Google Flights and Google Hotels, providing real-time information from Google’s partner companies, making it simple to compare both travel times and, importantly, costs.

This method provides a straightforward way to look for flights and lodging within your budget. For instance, I instructed Gemini to find me round trip flights from London to Paris for no more than £200. This serves as an excellent starting point to gauge your potential expenses and travel duration.

How to utilize it: Once you access Gemini (you might need to log into a Google account), open Settings and go to Extensions to ensure Google Flights & Hotels is activated. Then, return to the Gemini main page and input your request, detailing your departure and arrival locations, the duration of your visit, and any budget constraints you want to include.

If you love using spreadsheets, you can ask Gemini to export your itinerary to Sheets, which you can later share with family and friends.

Enhance your language abilities

You may have heard that practicing speaking is the best way to improve in a foreign language. However, hiring tutors can be costly, and you might not have anyone in your circle fluent in the language you’re aiming to enhance.

In September of the previous year, OpenAI upgraded ChatGPT to enable users to converse with it through speech. You can experience this for yourself by using the ChatGPT app available for Android or iOS. I opened the voice chat feature and recited some basic phrases in French, which it accurately translated into English (“Do you speak English?” “Can you help me?” and “Where is the museum?”) despite my lackluster pronunciation. It was also effective at providing alternative expressions when I requested less formal versions, such as replacing bonjour (hello) with salut, which means “hi.” Additionally, I was able to engage in basic dialogues with the AI voice.

How to use it: Download the ChatGPT application and tap on the headphone icon located beside the search bar. This will initiate a voice conversation with the AI model.

Translate while you’re out

Google has seamlessly integrated its robust translation technology into its camera features, allowing users to simply direct their phone camera at an unfamiliar phrase to see it converted to English. This is especially useful for understanding menus, road signs, and store names while exploring.

How to use it: Download the Google Translate application and select the Camera option.

Craft online reviews (and social media captions)
Positive feedback is an excellent way for small businesses to differentiate themselves from their online competitors. However, composing these reviews can be time-consuming, so why not utilize AI for assistance?

How to use it: By informing a chatbot like Gemini, Copilot, or ChatGPT about what you enjoyed regarding a specific restaurant, guided tour, or destination, you can simplify the process of writing a brief summary. The more detailed you are, the better the output will be. Prompt the model with something like: “Write a positive review for the Old Tavern in Mykonos, Greece, that mentions its delicious calamari.” While you may not want to use the response verbatim, it can certainly help with the structure and wording of your review.

Likewise, if you find it challenging to come up with captions for your travel-related Instagram posts, asking the same language models for help can be an effective way to overcome writer’s block.
November 4, 2024