The polymath ideal—embodied by figures like Leonardo da Vinci and Benjamin Franklin—was once attainable. During the Renaissance, a single curious mind could master multiple disciplines and make significant contributions across fields. These “Renaissance men” could reasonably grasp the breadth of human knowledge available in their time.
As humanity’s collective knowledge exploded post-Renaissance, people adapted by becoming increasingly specialized. Deep expertise in narrow domains became necessary to advance knowledge. The job market evolved to prioritize specialists—experts who could push boundaries in specific fields.
The ideal professional morphed into what we now call a “T-shaped” individual: someone with profound depth in one area (the vertical bar of the T) and a surface-level understanding of adjacent disciplines (the horizontal bar). This model recognized that while specialization was essential, some cross-disciplinary awareness remained valuable.
Despite this trend toward specialization, certain hybrid professionals continued to thrive: engineers with economic backgrounds, chemists with biological knowledge, or neurologists versed in computer science. These multidisciplinary thinkers often found unique insights at the intersection of domains, but they remained relatively rare exceptions to the specialist rule.
Now, large language models (LLMs) have changed the equation dramatically. These AI systems represent something unprecedented: a polymath in your pocket.
Modern LLMs encode such vast knowledge that they demonstrate PhD-level capabilities across numerous fields simultaneously. More remarkably, these systems are:
Widely accessible Rapidly improving Relatively affordable Available on demand
In my interactions with these systems, I’ve observed their remarkable performance when given precisely specified goals. They excel at detail-oriented tasks and can retrieve specialized knowledge across countless domains.
However, LLMs still have limitations. They’re not as adept at “big picture thinking.” While they can conceptualize ideas, they struggle to step back from details to question whether an approach is truly helpful or necessary. This oversight function remains a distinctly human strength.
The boundaries between human and artificial intelligence are already becoming well-integrated. As Ethan Mollick explains in his blog post “The Cybernetic Teammate”, AI can effectively complement human capabilities, leading to higher productivity and an improved emotional experience of work.
Given this new reality, I believe our educational ideal should shift. Rather than producing T-shaped individuals, we should aim for what might be called “rectangular” professionals—people who know numerous topics with substantial (though not complete) depth.
This breadth of knowledge would make them fully capable of:
Directing AI systems to handle detailed tasks Critically evaluating AI outputs across multiple domains Synthesizing insights across disciplinary boundaries
Such individuals would be extraordinarily productive. They would depend less on specialists in other fields, eliminating common friction points like:
Isolated knowledge silos Extended waiting times for expert input Hierarchical bottlenecks Interdepartmental conflicts More importantly, these AI-empowered generalists might recapture the Renaissance ideal: developing a comprehensive understanding of the world, pursuing creative endeavors across domains, and leveraging the unique perspectives that each discipline offers.
The polymath isn’t extinct—it’s being reborn through the partnership between curious human minds and artificial intelligence.