AI is rapidly evolving and impacting various aspects of contemporary life, but some specialists are concerned about its potential misuse and the impact on employment. AI is a technology that enables computers to imitate human actions and responses by processing large volumes of data to identify patterns, make predictions, solve problems, and learn from mistakes.
In addition to data, AI relies on algorithms, which are a sequence of rules that must be followed in order to carry out specific tasks. AI powers voice-based virtual assistants like Siri and Alexa and enables platforms such as Spotify, YouTube, and BBC iPlayer to suggest content. Furthermore, AI technology assists social media platforms like Facebook and Twitter in curating user content and supports companies like Amazon in analyzing consumer behavior to offer personalized recommendations and combat fake reviews.
Two popular AI-driven applications, ChatGPT and My AI Snapchat, are examples of “generative” AI. They utilize patterns and structures from extensive data sources to generate original content that simulates human creation. These apps are integrated with chatbots, allowing them to engage in text-based conversations, answer inquiries, weave narratives, and generate computer code. However, critics produce caution that these AI systems can erroneous responses and perpetuate biases present in the source material, such as gender and racial prejudices.
The absence of comprehensive regulations governing the use of AI has raised concerns about its rapid advancement. Some experts advocate for halting AI-related research, while others, including technology figureheads, emphasize the need for a rational discourse on AI’s capabilities. Notably, there are apprehensions regarding AI’s potential to propagate misinformation, influence societal decision-making, and even surpass human intelligence, leading to catastrophic consequences.
Governments worldwide are still grappling with the establishment of effective AI regulations. The European Parliament recently endorsed the European Union’s proposed Artificial Intelligence Act, which aims to impose strict legal guidelines for AI applications. The Act categorizes AI applications based on their potential risks to consumers, with varying levels of regulation.
Meanwhile, the UK has revealed its vision for AI’s governance, opting for oversight by a designated body rather than a dedicated regulator, while emphasizing the necessity for global cooperation in AI regulation. Additionally, China aims to mandate user notification of AI algorithm usage, reflecting the global discourse on AI governance.
AI has advanced to applications that can perform tasks previously requiring human intervention, such as customer interactions and gaming. While the term encompassing AI is often used interchangeably with subfields like machine learning and deep learning, it’s crucial to recognize the distinctions between these areas. For example, while all machine learning constitutes AI, not all AI incorporates machine learning. Many businesses are heavily investing in data science teams to fully harness AI’s potential. Data science integrates statistics, computer science, and business acumen to extract value from data.
Developers use AI to effectively perform tasks, interact with customers, recognize patterns, and solve problems. When beginning with AI, developers need to have a basic grasp of mathematics and be comfortable working with algorithms.
When starting an AI application development journey, it’s best to begin with a small project, like creating a simple application for a game such as tic-tac-toe. Practical learning can significantly improve any skill, including artificial intelligence. After successfully completing small projects , the potential for applying AI becomes limitless.
AI’s essence lies in emulating and exceeding human perception and response to the world. It is rapidly becoming the foundation of innovation. Fueled by various forms of machine learning that identify data patterns to enable predictions, AI can enhance business value by providing a deeper understanding of Abundant data and automating complex tasks.
AI technology improves enterprise performance and productivity by automating tasks that previously required human effort. It can also comprehend data on a scale beyond human capability, yielding substantial business benefits. For instance, machine learning has contributed to Netflix’s 25% customer base growth through personalized recommendations .
The adoption of AI is rising across various functions, businesses, and industries. It encompasses general and industry-specific applications, such as predicting customer spending based on transactional and demographic data, optimizing pricing according to customer behavior and preferences, and using image recognition to analyze medical images for potential illnesses.
According to the Harvard Business Review, enterprises primarily employ AI to identify and prevent security intrusions, address users’ technological issues, streamline production management, and oversee internal compliance with approved vendors.
The growth of AI across various industries is driven by three factors. Firstly, the accessibility of affordable, high-performance computing capability has significantly improved, mainly through cloud-based services. Secondly, abundant data is available for training AI models, made possible by Affordable storage, structured data processing, and data labeling. Finally, applying AI to business objectives is increasingly seen as a competitive advantage, leading to its prioritization and adoption across enterprises.
AI model training and development involves various stages, including training and inferencing. This process experimenting with machine learning models involves address specific problems, such as creating different AI models for computer vision tasks like object detection.
A few weeks back, I had lunch with a close friend who manages a rapidly growing real estate business with a $30 million annual revenue. While they primarily operate as a services business, he surprised me by discussing their extensive use of AI!
Their primary use case for AI is in customer service and support. With thousands of customers, they receive a substantial volume of messages ranging from support queries to feedback for improvement.
Initially, the company’s employees handled customer feedback. However, as the business grew, it became overwhelming. According to him, the critical challenge (and opportunity) was not just responding to people, but analyzing the feedback to gain actionable insights. This involved identifying themes for improvement or new features, services, or process enhancements.
Typically, such work is performed by a junior product manager. While not particularly challenging, historically, it required a human touch to interpret different comments (eg, “The food was sick!” and “The food was sickening!” represent two distinct types of feedback!)
AI came to the rescue. Instead of a human analyzing the data, he utilized AI for this task. He provided all the feedback and asked the AI to summarize, categorize, and recommend improvements and actions to take. This process took just a few minutes and was part of a twenty-dollar-a-month AI subscription!
Significantly, he found that Claude outperformed ChatGPT. The version of ChatGPT he used was a bit too “lazy”, often summarizing instead of categorizing everything, whereas Claude was more diligent in categorizing. Of course, this is a moment in time—OpenAI, Claude, Gemini, and others are continuously improving. Achieving the right balance between conciseness and accuracy versus wordiness and creating imaginary content has been a challenge for these AI platform vendors.
He also verified the AI results manually. Surprisingly, Claude’s results were actually superior to those done by an individual human.
Now, he is relying solely on AI to process the feedback, rather than hiring additional staff.
Another job lost to AI.
How many more jobs are in danger?
I suspect the actual impact will be even greater.
For any of my readers in a corporate or government position, consider how effective (or ineffective) your company is today—even without AI! Do you have any coworkers that leave you wondering, “What do they actually do?”
Having experience in both large companies and personally over the years, I have observed how inefficient organizations can be.
Bureaucracy leads to more bureaucracy!
Some companies have managed to combat encroaching bureaucracy. The changes made by Elon Musk at Twitter since he acquired it are remarkable. Set aside the political and media debate he has attracted and look at it from a business standpoint. He has now reduced the staff by around 80%, yet from an external standpoint, the company is thriving. New features are consistently being introduced (eg, subscriptions), and the service is still operational despite many critics predicting a complete collapse.
I delved deeper into the changes at Twitter last year on ThoughtfulBits. However, for this analysis, simply recognizing that inefficiencies exist in many organizations is sufficient.
At some point, at least one company in any industry will find out how to utilize AI technologies to eliminate or minimize those inefficiencies, providing them with a significant competitive advantage over traditional companies that don’t innovate.
So, is this the end? Will we see 30% or more unemployment in the upcoming years?
My personal prediction is no.
I make that prediction based on history. AI is not the first technological revolution the world has seen: farming, the industrial revolution, and the computer revolution, among others, have each dramatically transformed the job market.
In 1850, about 60% of the US population was involved in agriculture. Now, that figure is 3%. Historically speaking, food is now abundant and inexpensive. Although challenges regarding global poverty and hunger still exist, as a society, we have made tremendous advancements in food production while requiring far fewer individuals.
What happened to all of those farming jobs? They are now computer programmers and Instagram influencers. The idea that an Instagram influencer could be a legitimate profession was unimaginable in 1850 and controversial even thirty years ago! There are now millions of individuals working as influencers in an industry generating over $21 billion in revenue.
The World Economic Forum has some fascinating data on this shift over time.
I anticipate we’ll witness a similar shift as AI begins to take over entire job categories, particularly lower-level knowledge worker positions, as noted by McKinsey.
The Experienced Worker
The crucial question is: “What will these new jobs be?”
To answer that, let’s take a first principles approach: What remains constant in the world, even with AI?
Well, the first answer is people!! And everything people need to be happy fulfilled humans.
Even with AI, people will still need a place to live. They will still want to eat, go on dates, have families, play sports, learn, be entertained, socialize with friends, and so on. These are fundamental human and societal needs. While the context may be different, all those things were true in ancient Roman and Greek times, just as they are now. The Olympics originated in ancient Greece, after all!
With the rise of computers, we witnessed the emergence of the modern “knowledge worker” class—think of everyone working at an office for some company (as opposed to a factory or farm). These jobs, whether in digital marketing analysis or software programming and similar fields, emerged due to the computer revolution.
I expect we’ll see analogous “AI-focused” jobs. In fact, today, there is a new job category known as prompt engineering. Prompt engineering is for technical individuals focused on customizing AI technologies for specific use cases.
As a simple example, consider the questions you might ask ChatGPT—the better you frame the question, the better the results. This forms the core of prompt engineering. However, given how rapidly AI is evolving, it’s unclear how enduring the prompt engineering job might be.
Likewise, there will be numerous “AI consultants” in the upcoming years to assist individuals and organizations in transitioning to AI technologies, similar to the multitude of local “PC repair” shops in the 90s. But as people became more familiar with computers and the machines themselves became more reliable, those PC repair shops faded away.
Prompt engineers, AI consultants, and similar roles will proliferate for a period, but what jobs will be more steadfast and enduring in the post-AI era?
Returning to first principles, what is the common thread among most of those universal and timeless activities?
It’s about people interacting with other people.
If we extrapolate, just as the Industrial Revolution and the emergence of industrialized farming essentially opened up the economy for entirely new job categories, the replacement of many knowledge workers with AI will similarly create new opportunities.
I will categorize the new jobs after AI as “experience workers.” Some of these jobs we already know: tour guides, coaches, teachers, chefs, scuba divemasters, and more. For instance, consider dining at a fancy restaurant and watching the chef prepare your meal. This is an experience that cannot be replaced by AI or AI-controlled robots anytime soon.
While the nature of each of these jobs may be different, such as cooking versus scuba diving, they all involve human-to-human interaction and connection. This human connection is the timeless essence of being human.
In some cases, we might see an increase in the number of people in experience worker jobs. History offers insights into this. Industrialized agriculture has lowered food prices over time, leading to a rise in the restaurant business over the last century (consistently until Covid!).
Which jobs might see similar increases due to AI? Let’s consider teaching. While it’s easy to think that AI may reduce the need for teachers, tasks such as teaching a kindergartener to write require in-person interaction. AI can, however, make teachers more effective and efficient, handling tasks like grading and tutoring. This could lead to more teaching, not less.
For example, last winter, I tried Carv.ski, an AI and sensor package for snow skiing.
Using Carv was a fascinating and fun experience! Despite my thirty years of skiing experience, the AI considered my skills to be, well, “amateur at best”! It definitely helped me improve this season!
However, I still prefer an in-person ski instructor who can also access the data from the Carv system. That would be the best of both worlds – an instructor who can see how I perform in any snow condition, combined with the insights of the AI.
In essence, AI could make it easier and more cost-effective to be a ski instructor while improving outcomes. This combination can be powerful. Even without AI, many businesses, from FedEx to Shopify, have thrived by simplifying and reducing the cost of previously challenging endeavors.
This brief interview with the founder of Shopify is well worth reading! When Shopify started, the market for e-commerce software was tiny because it was so difficult to use! They made it easier, and now have over a million e-commerce stores on their platform.
AI tools will simplify and reduce the cost of numerous industries and scenarios.
Known Unknowns and Unknown Unknowns
Taking a cue from a famous quote by Donald Rumsfeld, the former Secretary of Defense, the really interesting question is: what are the jobs we don’t know about yet???!!
By definition, I don’t know what those are! But I believe the most interesting new jobs in the post-AI world will be ones that we can’t imagine yet, just as few people imagined the job of an Instagram influencer!
I also believe that these unknown jobs will involve people connecting with others in some way, as experience worker jobs do.
The Transition
I would be remiss not to comment on how quickly the changes in the job market may occur. As I mentioned at the beginning of this post, we are already seeing it, albeit in small ways (e.g., one less job posted in a startup). What if the job market changes happen really quickly?
It’s one thing to say, “Oh, there will be many more sports instructors, so no problem!” But it’s quite different when it affects specific individuals. If you’ve been laid off, that’s not a theoretical exercise. It’s a real, live “what do I do now and how do I support my family?” situation. It might be challenging to transition from an office job to a scuba or ski instructor or any newly invented experience worker job overnight, especially if you live in Kansas.
While I am hopeful that society will adapt to AI technologies, just as we have to every other technology revolution in history, the transition could be abrupt and messy.
That is a topic for another post, though!
In the meantime, if you’re working on AI, adopting AI, or are otherwise affected by AI, remember the importance of people! The relationships and social interactions between people are crucial. Technologies will evolve and enhance the human experience, but I don’t believe they will replace it. This is the opportunity for all of us!
The recent events involving tech CEO Elon Musk have brought him a lot of attention, particularly his acquisition of Twitter and the subsequent changes he initiated. Many people have been asking me about the significant reduction in staff, with some sources suggesting it’s been over 70%. This raises the question: is this truly achievable, let alone advisable? Could this lead to inevitable failure for him?
One Twitter user, Paul Vick (@panopticoncntrl), posted a tweet expressing that many tech CEOs seem to take delight in the fact that Elon let go of 75% of his workforce, yet Twitter is still functioning. However, the user believes that this situation might resemble the operations of Southwest Airlines, which could run smoothly until it encounters issues.
This tweet captures the prevailing sentiment on both sides of the debate. However, it fails to address the more crucial question: it’s not about whether you can downsize staff and keep the company functioning; the crucial question is, what problem are you attempting to solve?
As a former Chief Technology Officer at AOL, I have firsthand experience of implementing substantial staff cuts within a company. There’s no denying how difficult it was, especially for those directly affected. However, it was also a matter of survival for the company – we had to do it to stay afloat. And not only did the company survive, but many of AOL’s products remain active over a decade later.
Three essential forces are at play here: Customers, Employees, and Owners (sometimes represented by the CEO and senior executives). Each has a valid and compelling perspective.
From the employees’ standpoint, let’s consider that every job within a company is legitimate and valuable. Each employee was likely hired to fulfill a specific need and is currently engaged in meaningful work. Moreover, someone spent time, effort, and resources to secure their position. Another individual dedicated time to recruit and hire them. Someone is investing time in managing the employee. By and large, someone cares about that employee and their work. After all, how often do you talk to a friend working at a large company and hear them say, “Well, my job is pointless, and I have nothing to do”? Not very often.
This success leads to expansion, the hiring of more people, filling in skill gaps, and so on. There are a series of gradual improvements that go beyond the initial innovation. If you’ve ever had the chance to drive a luxury car like a Porsche, you can sense the decades of improvements in the driving experience.
Most of you probably use Microsoft Word. I doubt many of you would willingly go back to using Microsoft Word from 1995. The current version is a thoroughly refined and polished product. Yet if I asked you which single feature you couldn’t live without, you’d probably say “automatic spell check.” That feature was introduced in 1995!
Over time, it becomes easy to reach a point of diminishing returns on product refinement. These refinements are valuable to at least some set of customers—there’s typically a rigorous feature prioritization process! Yet these incremental refinements often lack the same impact as the original innovation.
A similar effect is observed with governments and government bureaucracy. As those of us in the United States prepare for our annual federal income tax exercise, we encounter the complexity of the tax code. Many of these regulations were introduced to address issues and special cases resulting from individuals attempting to reduce their taxes.
If you’ve ever had to complete government contracting forms, you’d have experienced a similar level of complexity. Even the number of pages, font, and font size are often stipulated.
Someone, somewhere in the past, undoubtedly attempted to submit an extensive proposal, leading to a rule about page length. Subsequently, another person used a small font, resulting in the rule on font size. There are over 2300 pages of rules for government contracting (and that’s just the baseline; the Department of Defense has an additional 1000 pages of supplementary regulations).
This iterative refinement works for a while until a disruptive change looms on the horizon.
This is where the customer dimension comes into play. It’s easy to perceive customers as a more uniform, homogeneous group, as seen in the countless business slogans: “Be customer focused. Customers are our number one priority. Customer-driven.”
However, as we all know, the reality is far more intricate. Some customers want no change at all, while others seek gradual improvements. Another group may desire more radical enhancements (in terms of cost, functionality, etc.). Even within those groups, there’s enormous diversity in opinions, desires, and needs. We used to say at Microsoft for many years: “No one uses 100% of the features of Office, but every feature is used by at least someone.”
The incremental planning and refinement process mentioned above is generally very effective at balancing the current customers’ needs. That’s why so many companies use it!
Managing disruptive change is the challenge. This kind of disruptive change may involve sacrificing some performance for cost, such as the original launch of gmail.com providing 1 gigabyte of storage when other email products offered 2MB—a 500:1 performance increase. At times, it introduces entirely new categories of functionality, like smartphones or AI and blockchain technologies in today’s world.
It may be challenging to accommodate diverse customer needs, especially when the disruptive technology would entail a significant change in the company.
In “The Innovator’s Dilemma,” Clayton Christensen delves into the difficulties successful firms encounter in adapting to new technologies or market shifts. I strongly suggest reading this book if you haven’t already.
Let’s take the case of Microsoft Word. I no longer utilize Microsoft Word—the transition was swift. Earlier, I would utilize Word on a daily basis; presently, I rely on chatGPT and Grammarly for all my writing tasks. The combination is remarkable: it has significantly enhanced both the speed and quality of my writing.
End-to-end software projects
The AI revolution encompasses more than just improving programming productivity—making the same activity more efficient. AI is also reshaping both the how and the what of numerous business processes. Building on the earlier example of outsourced programming, consider the full range of tasks involved in those projects.
An engineer typing on a keyboard and writing code is just one aspect. Additionally, there is project management, documentation, testing, regulatory compliance certification, user training, and more.
Some of these processes, such as regulatory compliance, can be extremely laborious and time-consuming. I have firsthand experience with a variety of compliance steps at different companies.
The legal department initiates the quarterly requests for a compliance update, which are then passed on to a group of compliance managers. They, in turn, approach different parts of the company for updates. In the case of compliance involving software, the compliance managers request updates from software program managers. These program managers then ask the engineers for the latest updates.
Needless to say, writing compliance reports is not the most enjoyable task for any engineer.
However, what if a compliance report could be generated at the click of a button? Moreover, what if the report also demonstrated to the engineers how to rectify the code to address those issues?
This would revolutionize compliance management. This capability would involve more than simply doing the same activity quicker. It would enable a complete rethink of the process and eliminate numerous hours of tedious work as it exists today.
Unquestionably, compliance is not the sole aspect of software development that is undergoing transformation. New AI developer tools can automatically document entire codebases and keep that documentation current. Tests can be automatically generated, and achieving the often-discussed “shift-left” cybersecurity objective (remedying cybersecurity issues in code rather than attempting to rectify them post-implementation) becomes significantly simpler with AI tools. The latest AI developer tools not only automatically identify cybersecurity bugs but also provide fixes to resolve the issues.
During the most recent earnings call, the CEO of Accenture, Julie Sweet, extensively discussed their work with legacy systems. Traditionally, this has been a source of competitive advantage for Accenture—they possess the teams and expertise to manage older and often outdated technologies. But what if AI tools could rewrite legacy software into more modern technologies?
These are not hypothetical scenarios. These AI-powered tools are currently available (full disclosure—my company Polyverse develops some of them!), and the tools are rapidly improving—sometimes on a weekly basis.
The leadership team at Accenture is certainly aware of these advancements in AI capabilities—Julie mentioned this in the aforementioned investor call, for instance. However, Accenture’s challenge lies in what action to take in response.
At present, Accenture talks a lot about AI but has yet to make any fundamental changes to their business.
Someone else will take the lead.
My forecast is that numerous smaller, more agile outsourcing firms will fully and vigorously embrace these new AI technologies. They will leverage these newfound capabilities to compete against Accenture and other “legacy” outsourcers.
However, these new proposals won’t just focus on pricing—they will encompass the complete package. An AI-enhanced outsourcing provider could offer better software delivered more rapidly, fully compliant, and better tested and documented, all at a significantly lower cost than legacy providers like Accenture.
In the beginning, these rivals will start by testing the waters. The proposals will appear too good to be true! Even though the proposal is accepted, enterprise sales will still be a time-consuming and lengthy process—so far, I haven’t witnessed any AI technologies that expedite the enterprise sales process!
At some stage, probably within a year, those initial attempts will evolve into a full-scale competitive rush.
Accenture and other major public companies will heavily publicize, promote, and make a fuss about their own implementation and embrace of AI.
Ultimately, they are constrained by their achievements. If staying competitive in the future means halving revenue, is it feasible for them? Can they acquire enough new customers and projects quickly enough to make up for the shortfall?
It’s not just a financial query. Culturally, these companies have a deep-seated emphasis on billable hours. If you are an employee there, that’s how you earn, receive bonuses, get promoted to management, and so on. Shifting that focus from billable hours to a “how do you accomplish this more quickly for less cost” mindset could be daunting.
Remember, this AI revolution is not simply about learning to use a new tool. AI is advancing at a rapid pace. In software development, last year, AI tools were essentially equivalent to advanced auto-complete. By the end of this past winter, they were capable of generating large sections of code. Now, the cutting-edge is complete code conversion, security testing, and compliance verification. Where will these tools be a year from now?
It’s not only AI programming that is rapidly progressing. In November 2022, ChatGPT 3.5 could surpass the legal threshold in the bottom 10%. By March 2023, ChatGPT 4.0 exceeded the threshold in the top 10%. Similar swift progress is being made in image and video generation, and so on. Where will we stand a year from now?
Providing value to customers as an AI-driven provider requires a completely different mindset than focusing on billable hours. It’s about continuously enhancing both efficiency and capability.
With Polyverse, we are fortunate to be collaborating with several partners who are fully embracing this new AI-driven mentality. There is a tangible sense of enthusiasm and determination—they all perceive billions of dollars of potential from established providers ready for disruption.
Leave a Reply