AI is transforming cognitive work across all areas of human activity.

It is poised to bifurcate humanity into superhumans who are amplified by it and those who become disempowered.

Individuals will be challenged to see if they can stay on the right side of the divide.

AI is now writing its own code and learning increasingly by self-supervision. It’s like an intelligent alien species is now occupying the planet with us and is becoming smarter by the day as it learns about us. How will we live alongside this machine in the future?

One of the major themes in my book, Thinking With Machines, is that AI is poised to bifurcate humanity. It will amplify those with deep knowledge about something while disempowering others who don’t. As a “pracademic” who brought machine learning to Wall Street in the 90s, AI amplified my edge over those who were not aware of its potential. I have continued to use AI to create novel commercial trading strategies since that time.

But in the new era of modern AI, I am forced to consider whether I am becoming obsolete, both as a practitioner and professor. Many of us should be asking ourselves the same question: how do I stay ahead of the machine?

Let’s construct a scoreboard for and against my potential obsolescence, starting with the practice side of my pracademic existence. My latest investment strategy took a couple of years from concept to readiness, and another year of “paper trading” by my client to get comfortable with the strategy. The research required a deep understanding of both markets and cutting-edge AI algorithms. However, my multi-year effort – which coincided roughly with the emergence of LLMs – can now be accomplished in a few days by a handful of talented data scientists – as long as they know the right questions to ask and interpret the outputs of the AI correctly along the way.

That’s not a trivial qualification. After all, it took decades of experience in markets and AI to develop an ability to ask the right questions and make sense of the intermediate results. Experience has also taught me humility and an appreciation of the inherent uncertainties and unpredictable shocks that characterize financial markets. Experience teaches us to recognize when something doesn’t feel right and to sense when a path is likely to lead to frustration and dead ends, or worse still, to flawed models that perform badly in practice.

The question is whether AI can compensate for a person’s lack of experience by enabling anyone sufficiently curious to formulate/conceive the right questions. Can the AI learn to recognize why a result feels off? Or why an approach is likely to be a dead-end even if the theory and numbers look good? Can it learn such things vicariously from the experience of scores of professionals like myself? Is it a matter of time before the AI acquires sufficient curiosity and experience to render my edge obsolete? I think the answer to all these questions is a “yes.” AI will learn all these things. It’s sobering to realize that this is just a matter of time.

On the teaching front, I encourage my students to use AI to amplify their critical thinking skills, and to explore new ideas using the AI as a sounding board and critic. In the classroom, however, I rely on my experience to channel class discussion, and find that many of the most valuable insights occur spontaneously. Last week, a student told me that he turned to ChatGPT when he didn’t follow something I said, which enabled him to get on track in real-time. Another student told me that he fact-checked something that didn’t accord with his intuition while I was speaking. It made me wonder how far we are from an AI digital twin of me that is able to guide real-time discussion in interesting ways, that is customizable for the audience. At the end of every class session, the AI could also figure out how to best segue to the next session!

The AI might also do a great job at evaluating students and guiding them in a personalized way based on individualized assessments. For example, it could assess how candidates would hold up in a professional discussion and coach them accordingly. Imagine asking your personalized AI teacher something like, “Perform an interview for the role of an equity research analyst as if you were Professor Damodaran.”

AI amplifies the abilities of some while disempowering others. People who already know a lot are in a good position to know when the machine is wrong or needs a course correction, and to be able to nudge the machine in productive directions.

The challenge is how to stay on the right side of this impending bifurcation. I assumed that I’d fall on the right side of this divide. After all, I’ve been teaching and practicing AI for decades, so I have developed skills honed by trial and error. But now, I ponder how long such skills will be an edge as AI advances, and how I will add value in the future. One thing seems clear: I can’t let the machine do my thinking for me. Rather, I must up my game so that I get better at thinking with AI. This is one of the biggest challenges for all of us in the years ahead.

How does it apply to you?

© Psychology Today