What does Tim Gowers disagree with in the Fields medallists' letter?

Gowers rejects the letter's ranking of mathematical goals. The declaration says solving problems is only a tool and proxy for the real aim of conceptual understanding; Gowers argues mathematics holds a spectrum of temperaments, with some people driven mainly by problems and others by understanding, and that calling one a mere proxy for the other delegitimises half the field.

Does Tim Gowers want AI labs to slow down?

No. Gowers explicitly declines to recommend slowing AI development, treating it as futile, and that is part of why he did not sign. He reads the letter as aiming blame at AI companies when the pressure he worries about sits inside academia — the incentives that decide whether people still train as mathematicians at all.

What does Gowers think the real crisis in mathematics is?

Loss of human motivation, not an unmanageable flood of results. Gowers writes that the risk is that a lot of people who would have done a PhD in mathematics will no longer wish to do so once machines solve the famous problems, which would break what he calls the crucial human transmission chain that keeps mathematical culture alive.

What does Gowers suggest mathematicians actually do?

Gowers asks PhD supervisors to find imaginative new projects for students, wants proper writeups and citation of prior work enforced now — using AI itself to find the work being reused — and floats databases that catalogue AI-produced results. He also argues the community urgently needs a way to explain to funders why a large pool of human experts is worth paying for.

Does Gowers have a financial tie to an AI lab?

Gowers discloses in the post that he is given access to OpenAI models but says he has never received payment. He also writes that he has swallowed the bitter lesson: AI now solves problems without the human-style insights his Cambridge automated-theorem-proving group spent years pursuing, which shapes his view that resisting the technology is not the answer.

Tim Gowers · 2026-09-17 · major

Tim Gowers — why he didn't sign the Fields medallists' AI letter

Tim Gowers explains why he stayed off the declaration signed by 25 Fields medallists. He agrees mathematics faces a crisis but names a different one: too few people will want to start a maths PhD once AI can solve the famous problems.

Tim Gowers speaking at the GOSIM conference at Station F in Paris — Wikimedia Commons

Gowers agrees mathematics is in trouble, but thinks the letter blames AI labs for a problem that lives inside academia.

Quick facts

Author	Tim Gowers
Published	17 September 2026
Responding to	The declaration signed by 25 Fields medallists
His main worry	Fewer people will start a mathematics PhD
On slowing AI down	Does not recommend it — calls it futile
Disclosure	Has OpenAI access, says he has never been paid

What is it?

Tim Gowers published 'Why I didn't sign the Fields medallists' letter' on 17 September 2026, six days after 25 Fields medallists signed a declaration asking AI companies to hold back on releasing models that crack famous open problems and to publish proper writeups that credit prior work. Gowers says he agrees with much of the text but could not sign confident recommendations he is unsure about.

How does it work?

The disagreement is about what mathematics is for. The declaration treats problem-solving as a proxy for conceptual understanding; Gowers answers that the field has always held both temperaments and that ranking them writes off the problem-solvers. He also pushes back on the idea that AI proofs appear from nowhere, noting they usually sit on a large amount of human mathematics, with weak writeups as a temporary annoyance rather than the core harm.

Why does it matter?

This is the first substantial public dissent from inside the same group of mathematicians, and it moves the argument from 'should labs slow down' to 'will anyone still train as a mathematician'. Gowers's practical asks — reimagined PhD projects, enforced citation of reused work, a case to funders for keeping human experts — are things departments and labs can act on without waiting for AI companies to agree to anything.

Who is it for?

mathematicians and AI researchers following the maths dispute

Frequently asked questions

What does Tim Gowers disagree with in the Fields medallists' letter?: Gowers rejects the letter's ranking of mathematical goals. The declaration says solving problems is only a tool and proxy for the real aim of conceptual understanding; Gowers argues mathematics holds a spectrum of temperaments, with some people driven mainly by problems and others by understanding, and that calling one a mere proxy for the other delegitimises half the field.
Does Tim Gowers want AI labs to slow down?: No. Gowers explicitly declines to recommend slowing AI development, treating it as futile, and that is part of why he did not sign. He reads the letter as aiming blame at AI companies when the pressure he worries about sits inside academia — the incentives that decide whether people still train as mathematicians at all.
What does Gowers think the real crisis in mathematics is?: Loss of human motivation, not an unmanageable flood of results. Gowers writes that the risk is that a lot of people who would have done a PhD in mathematics will no longer wish to do so once machines solve the famous problems, which would break what he calls the crucial human transmission chain that keeps mathematical culture alive.
What does Gowers suggest mathematicians actually do?: Gowers asks PhD supervisors to find imaginative new projects for students, wants proper writeups and citation of prior work enforced now — using AI itself to find the work being reused — and floats databases that catalogue AI-produced results. He also argues the community urgently needs a way to explain to funders why a large pool of human experts is worth paying for.
Does Gowers have a financial tie to an AI lab?: Gowers discloses in the post that he is given access to OpenAI models but says he has never received payment. He also writes that he has swallowed the bitter lesson: AI now solves problems without the human-style insights his Cambridge automated-theorem-proving group spent years pursuing, which shapes his view that resisting the technology is not the answer.