This is a piece I originally posted to my Instagram story in February 2026.

In an interview, NVIDIA CEO Jensen Huang was asked, “Who is the smartest person you’ve ever met?” Having spent his career alongside countless geniuses who reshaped the world, everyone expected him to name a specific person. But his answer caught them off guard. Instead of naming anyone, he questioned the very “standard of intelligence” we’ve long held sacred.

He pointed out that the domains once considered the summit of intelligence — intricate programming, high-level technical reasoning — used to be the exclusive territory of a handful of experts with sky-high IQs and years of training. Yet now, artificial intelligence solves these very technical puzzles faster and more accurately than they can. In other words, the era in which technical proficiency was synonymous with how intelligent you were has ended, and we therefore need a new definition of “smart.”

He boiled down this new kind of intelligence to three traits. First, being technically sharp while also understanding people. Second, being able to imagine what was left unsaid and what cannot be known. Third, being able to anticipate what the future holds.

This shift calls to mind a famous anecdote about the greatest genius of the twentieth century, Albert Einstein. We tend to remember him as an omnipotent mathematical genius, but Einstein himself liked to say, “I have no special talent for mathematics” (relative to his colleagues, at least). In fact, while developing the theory of general relativity, he had to lean on his friend, the mathematician Marcel Grossmann, to work out the complex non-Euclidean geometry involved. But what matters is this: the revolutionary insight that spacetime itself could curve came not from Grossmann the mathematician, but from Einstein, who grasped the essence through intuition.

The changes unfolding in mathematics and science today are extending this division of labor that Einstein experienced — only deeper, and on a global scale. Reinforcement-learning systems solve International Mathematical Olympiad problems, AI predicts protein structures and finds solutions to complex physical equations. The mathematical craft of computation and formal reasoning is rapidly migrating into the domain of an enormous machine collaborator.

“Imagination is more important than knowledge. Knowledge is limited, whereas imagination embraces the entire world.” — Albert Einstein

What emerges at this juncture is not a shrinking of human intellect, but a shift of its center. Just as Einstein had Grossmann as a collaborator, we now have AI as a high-performance calculator. Technical ability is no longer simply the capacity to compute well; it is reconstituted as the intuition to understand how machines work, to grasp their limits, and to discern which of the countless results AI produces is a “theorem that genuinely matters to humans.”

The insight to decide what is worth computing in the first place still belongs to humans. AI can search for patterns, but judging which pattern explains the world more deeply — and what value that discovery holds for humanity — requires human context and empathy. Just as Einstein agonized as he watched his theory lead to the atomic bomb, the choice to weigh how a technology’s application will affect society, and to take responsibility amid uncertainty, is hard to leave to AI alone (for now, at least).

And so, in the age of AI, the definition of the “smartest person” changes. He is no longer the one with the highest IQ or the best memory. Those abilities have already become ordinary, shared with machines. The genius of the new era is someone who combines a sharp intuition for “what matters” while riding the wave of technology, a sense for the lives of others, and the judgment to set a direction amid uncertainty.

In an age where AI writes proofs and proposes scientific hypotheses, human intellect is evaluated on a different plane. Under the new standard, the smartest person might well get a terrible score on a college entrance exam. But he is not someone who has memorized more answers; he is someone who knows what the “most valuable question” to ask an AI is, and who can bear the weight of the consequences that question brings. Perhaps the reason Jensen Huang challenged the definition rather than naming a person is precisely that he had witnessed this seismic shift in the intellectual paradigm.