Terence Tao, the UCLA mathematician widely regarded as the most influential living figure in the field, has published an essay arguing that artificial intelligence is pushing mathematics into a period of turbulence comparable to the foundational crisis that reshaped the discipline a century ago. The paper, "Mathematics in the age of AI," appeared on arXiv on August 17 and is based on a public lecture Tao delivered at the 2026 International Congress of Mathematicians.
The essay has quickly become one of the most discussed scientific texts of the week among researchers and engineers, drawing large audiences on forums like Hacker News. For a field that has spent the past three years debating whether AI can really do research-level mathematics, Tao's argument changes the question entirely.
A Crisis of Values, Not Truth
Tao's historical framing is deliberate. Between roughly 1900 and 1930, mathematics was shaken by Russell's paradox in 1901 and Gödel's incompleteness theorems in 1931 — discoveries that forced mathematicians to abandon informal assumptions and build the explicit, rigorous foundations the subject still rests on today. That turbulence, he argues, ultimately made mathematics stronger.
The parallel he draws is precise: what is being stress-tested now is not mathematics' framework for truth, but its "largely implicit framework of mathematical values and practices" — what the community considers a contribution to be, what it rewards, what it regards as understood, and who, or what, it regards as having done the work. Just as the last crisis forced the implicit to become explicit, Tao argues this one will force mathematicians to codify goals that have never had to be written down.
Skipping the Capability Debate
The essay's most striking structural choice is that it refuses to argue about whether AI can do mathematics. Instead, Tao conditions on the hypothesis that AI tools capable of research-level mathematical tasks will arrive, and treats everything downstream of that assumption as the real problem. He calls this the "Community Response Question": how should the mathematical community respond, not whether the capabilities are real.
It is a deliberately orthogonal move. Debates over model capability, in Tao's telling, are both unresolvable in the abstract and beside the point. The urgent questions — what mathematics is for, and what its practitioners actually value — would demand answers even if AI progress stalled tomorrow.
Proof Scarcity, Proof Abundance, and Goodhart's Law
Central to the essay is a case study on problem-solving, the component of mathematical practice most directly threatened by automation. Tao examines what happens when the field moves from an economy of proof scarcity — where verified, original proofs are rare and precious — to one of proof abundance, where machine-generated arguments are cheap and plentiful.
His list of keywords signals the ground he covers: formalization, proof assistants, peer review, Goodhart's law. That last item — the principle that a measure ceases to be a good measure once it becomes a target — recurs throughout. Publication counts, competition results, and theorem provenance have all served as proxies for mathematical contribution. In a world where AI can inflate those proxies at scale, the community will need to reward what it actually values rather than what it has historically been able to count.
The essay closes with recommendations and a survey of emerging workflows — including formal proof environments of the kind Tao himself has championed in experiments with AI-assisted, formally verified mathematics — alongside a candid acknowledgment that the community's answer will evolve as the technology does.
Why a Mathematics Essay Matters Outside Mathematics
Tao is an unusual figure: a Fields Medalist who has spent years working directly with proof assistants and AI tools rather than commenting on them from a distance. That practical engagement gives the essay weight beyond pure mathematics, because nearly every research discipline faces the same inversion he describes. Peer review, authorship, credit, and the definition of a contribution are under the same pressure in physics, biology, and computer science — fields that lack mathematics' century-old head start on making their foundations explicit.
The reception so far suggests the argument has landed. Discussion threads have focused on Tao's scarcity-to-abundance framing as a useful lens for academia at large, with particular attention to his point that the crisis of values, unlike the crisis of foundations, cannot be resolved by a small group of logicians. It will have to be worked out by the entire community — journals, hiring committees, grant bodies, and the mathematicians whose careers are built on the implicit norms now under strain.
Tao's conclusion is ultimately constructive rather than alarmed. The foundational crisis produced a trusted framework that survived a century of testing. A thorough, honest examination of what mathematics actually values, he argues, could leave the discipline stronger and more resilient than before — provided the community does the work of making its unwritten rules explicit before circumstances make the choice for it.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
