Humanity’s Big, Wrong Assumption About AI

Was AI ever meant to level the playing field? Or did we just assume it would? Early research was promising, with AI (especially Generative AI) being seen as a tool that would enhance productivity for everyone. It made marginal writers better, enhanced the quality of routine legal tasks, and made novice customer-support agents 34% more productive, with faster and more effective query resolution. And we embraced that research, framing AI as something that would increase productivity, automate tedious tasks, and unlock human potential. AI was prophesied to benefit lower-skilled workers by giving them access to the same tools and insights as the experts. Business leaders like NVIDIA’s Jensen Huang and OpenAI’s Sam Altman painted (and continue to paint) AI as a tool that would make everyone more capable. It has. Lower-skilled workers are more capable. However, highly skilled workers are rendered much, much more capable.with AI assistance. AI isn’t a democratiser. It’s a cognitive amplifier. No one ever guaranteed that AI would create a fairer system. That was our hopeful assumption,not a built-in feature of the technology. So what’s happening? AI looked like an equalizer at first. It made entry-level work easier, gave novice employees a boost, and saved time on tedious tasks. Then something else happened. AI started taking over those tasks. Now, instead of lifting everyone up, it’s making the gap between weak and strong thinkers impossible to ignore. AI is holding a mirror to humanity and exposing something that has always been true: Some people are great at thinking, and some aren’t. AI makes that distinction painfully obvious in a way that previous technologies haven’t. The initial divide is between those who are using AI and those who aren’t. As usage increases, this divide changes. The new divide lies between those who use AI well by thinking critically, strategically, and adaptively and those who rely on AI without engaging their own deeper reasoning, and therefore struggle to apply it effectively. In short, the divide is between those who use AI to help them think, and those who use AI to do their thinking for them. A recent article in the Economist (linked below) provides detail. AI initially helped novice workers to perform tasks. It’s now starting to take over those tasks with AI automation. On the other hand, experts who can refine and critically evaluate AI outputs are rewarded with better strategic decisions and insights. In fields like finance, law, and research, AI isn’t closing the gap between lower and higher skill levels. It’s expanding it. What are the thinking skills that set these experts apart? Critical Thinking:questioning assumptions, evaluating evidence, and distinguishing between sound reasoning and fallacies. Interrogating ideas for validity, coherence, and relevance. Independent Thinking: Forming independent conclusions rather than absorbing the beliefs of others. Resisting groupthink and intellectual laziness. Creative Thinking: Making unexpected connections, considering alternative perspectives, and generating new ideas. The ability to see problems differently. Systematic Thinking: Seeing the big picture, recognising how different elements of a system interact, and understanding cause and effect beyond the obvious. Reflective Thinking: Turning the lens inwards to examine one’s own biases, cognitive habits, and emotional reactions. Philosophic Thinking: Questioning fundamental concepts (like truth, justin, meaning, and existence) rather than focusing on their practical application. These thinking skills form even more of an advantage than pre-AI. The ability to think well, and to demonstrate these skills are now a human USP. What about those who can’t keep up? What happens to those people who can’t improve their thinking skills in an AI-driven world? These skills aren’t easy to develop, but thinking is a skill that can be improved. How do we teach it - and to whom? Despite the best teaching, though, everyone has a natural limitation on their cognitive function. Their thinking ability. So, what happens to those who naturally have more limited thinking skills and function than others? AI is already swallowing up the jobs that used to go to weaker thinkers. So what happens next? Do we redesign work to accommodate them? or do we just accept that some people won’t keep up? When machines took over factory work, people moved into offices. When software automated bookkeeping, new jobs popped up in tech. But what happens when AI starts replacing human judgment? Where do people go when thinking itself is the thing being automated? Do we accept that AI is sorting people into winners and losers? Are there ways to democratise thinking itself? There aren’t any hard answers at this point. AI is evolving at a tremendous rate, and even more of the benefits and drawbacks of this technology are going to come to light as use cases increase. Some feel that AI will make our lives unrecognisable in less than a decade. It’s interesting that this is often framed as the creation of an AI-driven utopia rather than a more sinister transformation. Perhaps we need to more careful about what we consider to be true of AI. Perhaps we need to think a bit more deeply about whether we’re thinking well enough to survive the world it’s building. First posted on LinkedIn on 16 February 2025.

The Economist. (2025, February 13). How AI will divide the best from the rest. Retrieved from https://www.economist.com/finance-and-economics/2025/02/13/how-ai-will-divide-the-best-from-the-rest Customer Support Productivity Study Brynjolfsson, E., Li, D., & Raymond, L. (2023). Generative AI at Work: The Impact of AI on Productivity and Workflows in Customer Service. Stanford University & Massachusetts Institute of Technology. Retrieved from https://papers.ssrn.com/ Writing Quality Improvement Study Noy, S., & Zhang, W. (2023). Experimental Evidence on the Productivity Effects of Generative Artificial Intelligence. Massachusetts Institute of Technology. Retrieved from https://www.nber.org/papers/w31161 Legal Work Enhancement Study Choi, J. P., Hickman, K., Monahan, A., & Schwarcz, D. (2023). ChatGPT Goes to Law School. University of Southern California & University of Minnesota. Retrieved from https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4335905