Start here

100+ reactions to 100+ solutions

The hosts of Proofs and Prompts · 8 October 2026

Mathematicians’ first reactions to OpenAI’s release, on 6 October, of solutions to hundreds of open problems, collected between 8 and 19 October. It is long, but each reaction stands on its own.

From non-sofic to non-hyperlinear groups: the new messiness of discovering and announcing big results

Francesco Fournier-Facio · 25 September 2026

An account from inside group theory of three long-open problems that fell to AI systems within two months, and of what OpenAI’s announcement said, and did not say, about the earlier human work it built on.

Open Letter about the Mathathon

A coalition of current and former Caltech mathematicians · 10 September 2026

Objections to a hackathon in which teams would prompt LLMs to solve open research problems. Read it with the organisers’ response, published the same day, and the joint statement of 28 September, in which both sides redesigned the event around new proofs of old problems.

Care for a little more AI?

Hugo Duminil-Copin · 30 August 2026

Through his own long history with one conjecture, argues that AI as it is used today freezes mathematical activity instead of empowering it, and explains why he does not rely on it for his own creative work.

A beginning for mathematics

Daniel Litt · 14 September 2026

Assuming that AI will soon be better than humans at most of mathematics, sets out what the field is trying to produce, both mathematics and mathematicians, and proposes concrete steps.

The AI dissenter viewpoint

Tasmin Chu · 9 August 2026

The fullest case on the site against taking up LLMs in the profession, centred on the social conditions that sustain human understanding. Two replies to it are linked at the end of the post.

Where to Point the Telescope

Shaowu Zhang · 21 August 2026

How AI workflows helped with the computations behind a result on the inverse Galois problem, which decisions stayed with the human collaborators, and how the final claims were verified without AI.

Is There Actually a Good Way to Use AI for Math?

Yushang Cheng · 22 October 2026

An undergraduate describes a way of working with AI, dropping assumptions, looking for counterexamples and tracking what each condition contributes, shown on a simple example. The author worries it can turn research into pressing buttons, and asks how experienced mathematicians work.

Writing mathematics in the age of AI

Martin Hairer · 7 August 2026

The site’s first post and its most practical: understand every argument before using it, disclose meaningful AI contributions specifically, and trace where an apparently new idea came from.

YAPOAI (Yet Another Post on AI): AI, understanding, and mathematical work

Najib Idrissi · 3 October 2026

A frequent user on what the models do well and badly, why checking their output is still real work, and whom a theorem belongs to.

Who will become the new mathematicians of Math 2.0?

Anonymous, PhD student · 20 October 2026

A student who has decided to leave academia after the doctorate asks senior mathematicians how a “Math 2.0” future would train the next generation, and whom it would attract. Read it with Álvaro Lozano-Robledo’s reply of 23 October.

Navigating the AI crisis: a humble guide for students

Tarik Aougab · 23 September 2026

Advice for students that deliberately neither pushes AI nor rules it out: divide your time between the fundamentals and what may become the new fundamentals, and invest in exposition and communication.

Recommendations on the use of AI for PhD students and supervisors

A collective of symplectic geometry researchers · 19 October 2026

Numbered recommendations for doctoral students, for supervisors and for universities, starting from the principle that mathematicians, not AI, decide which questions matter and judge the results.

Dear conferences, journals, and hiring committees: don’t wait for everyone else

Amit Harlev · 25 September 2026

Argues that hiring committees, journals and conferences should say now which contributions they will value, pointing to the rise in arXiv submissions as the start of a publication crisis.

Where will all the papers go?

David Craven · 19 September 2026

Argues that journals and referees cannot absorb the rise in submissions, and proposes that journals refuse AI-generated papers.

The Rules Cannot Change Only for Young Mathematicians: AI and the future of a mathematical career

Anonymous, Postdoc · 22 October 2026

On priority, disclosure, unequal access to models and the training of PhD students, ending with the questions the author wants hiring committees, journals and supervisors to answer publicly.