Terence Tao

Terence Tao (born 1975 in Adelaide) was a child prodigy who became a professor at UCLA at 24 and won the Fields Medal in 2006. Working across harmonic analysis, number theory, combinatorics, and PDEs, he is routinely described as the greatest mathematician of his generation — which makes his embrace of AI collaboration one of the most consequential endorsements in science.

The co-pilot experiments

Tao does not speculate about AI in mathematics; he runs the experiments and publishes the lab notes on Mathstodon. He led the crowdsourced formalization of his own PFR-conjecture proof in the Lean proof assistant, uses GPT models to draft and translate proofs into formal code, and served on the advisory committee of the AI Mathematical Olympiad. Testing OpenAI’s o1 model in September 2024, he compared advising it to advising a mediocre but not incompetent graduate student — then carefully clarified that real graduate students learn, improve, and out-create any static model.

In a 2024 Scientific American interview he predicted AI would become a genuine co-pilot for mathematicians within about three years, and in The Atlantic that October he sketched a future of “industrial-scale mathematics”: broad, crowdsourced problems attacked by teams of humans and AIs, with proof assistants guaranteeing correctness so collaborators no longer need to trust each other line by line. AI, he noted, is good at distilling billions of data points into one answer; humans are good at inspired guesses from ten observations — complementary strengths, not competition.

Tao’s stance is meticulous rather than boosterish: he documents failures as readily as successes, and his measured, evidence-first enthusiasm is precisely what gives it force.