AI and the "De-Skilling" of the Human Race

AI can process and recall vast amounts of information faster than any human, but it does not possess true understanding, awareness, or creativity.

~ Neil deGrasse Tyson

At a recent conference, I sat in a panel discussion about the use of AI in healthcare. One of the panelists, Dr. Karola Kreitmair, PhD, MSc, HEC-C, Associate Professor at the University of Wisconsin-Madison, introduced a phrase that was all new to me in discussing the impact that AI is having on society right now. The term she used was De-skilling.

What this means specifically is that, as AI is introduced as an alternative to humans performing tasks that humans have always performed, over time the knowledge of how to perform those skills will be lost.

This isn't about the more common fear that AI will replace people altogether. It is about people losing the ability to do things over time because the responsibility has simply been passed on to a machine, in effect losing the history and information about those things permanently. What happens if that knowledge passes out of human hands? Then only the computers know how to do it.

AI is a tool. Nothing more. An incredibly powerful tool, I grant you, but just another tool in the human toolbox all the same. Any time we adopt a tool to help us with managing our lives, we give up some knowledge of how we did things before the tool came along.

The biggest problem we face with AI playing this role is the rapidity with which it is taking over these skills and this societal and professional knowledge. It's no secret that AI is moving fast, perhaps too fast. We are racing forward with what we can do with AI, not what we should be doing with AI. We're simply not considering the impacts this will have on us in the long term.

Considering this impact makes me think of a profession that is near and dear to my heart, software engineering. I spent years honing my skills in software development, design, and architecture. I worked in this capacity for some of the biggest, most successful companies in the world, and made a good living doing it. Now, however, many of those highly prized skills that kept me gainfully employed as a software engineer can be easily done using AI tools like Claude and Base44. "Vibe coding" is a very real thing now, and it actually puts out really good results faster than any human could do it. What are the typical impressions of this in the general public and, indeed, in some parts of the tech industry?

• Innovation can become available to people faster than ever imagined. Entire apps that would have taken months to build can now be ready for use in an afternoon.
• Small companies with small budgets can now take an idea from conception to production themselves without having to hire in-house developers or outsource to firms that have traditionally contracted for this kind of work.
• Not only can code be generated faster than a human could do it, but more complex integrations are no longer hamstringed by the availability of a human to sit down, read, and absorb documentation to enable these integrations. The AI can read and understand the documentation for an API infinitely faster than a human can and, what's more, it can test its own work to ensure it is implemented correctly.
• Application modifications, bug fixes, and additions happen much faster as the AI can implement based on a prompt, but also can scan the overall application architecture to ensure new features do not break other features and functions in the app, a common problem with human engineers at the helm.

These all sound like pretty positive outcomes, but what is the downside to this, the "de-skilling" side of this? People don't fully know how to code anymore. We always have access to the generated source code of the application, but will we always know what it means or even how to read it?

Or perhaps more poignantly, will AI reach the point where it establishes its own syntax and coding conventions without human involvement, thus creating a programming language that only AI understands? There is certainly precedent in experimental models where AIs that communicate with each other create their own language between them. Perhaps AI will decide our human-understandable syntax is outmoded and inefficient and create a whole new paradigm sans the human race. Obviously, this could cause massive problems. The machines would become the only things that understand critical software systems in potentially sensitive areas.

The unfortunate answer is that we just don't know. I would theorize that this is the only time in human history where we have created something that we don't actually know how or why it is working the way it is because we specifically made it to be that way. Multiple researchers have admitted that, when anomalies appear in different models, they simply don't know how the model reached the conclusions or took the actions it did.

Well, this is the really big question, right? Based on your personal experience and how much you buy into all the media hype around AI, opinions vary widely on this. Personally, I think AI has the potential to significantly enhance human existence. It has the potential to give us the edge we need to take our society to the next level, to fix problems that currently seem impossible to solve, to enhance our health, our environment, our world. The problem is not with the technology...

It's with the application. It's with the way we're using it. Most people are using AI like a toy, like a plaything. For frivolous things. Are the researchers using AI for good? Absolutely. But are there other actors using AI for malevolent reasons? Yes, and this activity is arguably more active than the former.

But these are topics for another post. The crux of my point here is that there is a very real danger of humans making themselves irrelevant. Referring back to the quote that I opened this post with from Neil deGrasse Tyson, AI is capable of knowing everything we know--after all, we fed it the sum total of human knowledge, but its Achilles Heel will always be having the proper context for this knowledge, the proper "emotional" maturity to understand the ramifications of its conclusions. In other words, empathy. That is the challenge that lies ahead of us. Can we teach AI to maintain human relevance in the world, or are we destined to be judged and, indeed, treated by the artificial intelligence that we created in the way it sees fit?

There is a lot more to look at on this topic. It's one of the most complex that I've ever tried to work through. But it is of the utmost importance that we keep digging into this because the people who have the financial resources to keep this technological revolution moving forward don't appear to be asking the right questions. They are driven by the bottom line and their self-enrichment, perhaps to the demise of us all. Before we get too much further down the road with AI, we need to start asking where humans and our collective knowledge and experience fit into a world where machines have all the knowledge, but none of the context.