🔍 Read the full analysis: OpenAI’s AI Mathematics Has 722 Proofs. What Could They Lead To? on ThorstenMeyerAI.com
Get school and study supplies delivered free — and shop member deals
- Fast, free delivery on millions of items
- Access to Prime Big Deal Days deals on October 6–7
- Prime Video, Amazon Music and more included
TL;DR
OpenAI has published 722 mathematical manuscripts, organized into 372 families, from work by an unnamed, unreleased model. The collection includes claims about long-standing problems, but many results have not been formally checked, and the company warns that some may contain issues. Whether the work produces lasting mathematical advances will depend on independent verification and whether researchers can extract reusable ideas.
OpenAI published 722 mathematical manuscripts on Monday, generated by an unnamed model that the company has not released. The papers are grouped into 372 families of related results and include claims about major open problems, but OpenAI chief executive Sam Altman said the claims have not yet been confirmed by outside mathematicians.
OpenAI’s post and accompanying GitHub repository describe work across number theory, geometry, operator algebras, topology, theoretical computer science and mathematical physics. The collection came from about 4,000 problems posed to the model; OpenAI selected those it considered significant. The company says the average result took about three hours of ChatGPT Pro thinking compute. The problem selection and significance screening were conducted by OpenAI, not independent mathematicians.
The manuscripts include claimed results on the Unique Games Conjecture, Hilbert’s tenth problem over the rationals, the isomorphism of nonabelian free group factors, and the Hodge conjecture for certain abelian varieties. Another paper claims a zero-free region for the Riemann zeta function to the right of Re(s) = 11/12. These are claims in the released manuscripts, not findings established by outside review.
OpenAI says many, but not all, results have Lean formalizations, which can allow a proof to be checked by proof-assistant software. Its repository warns that “some of the unformalized results could have issues.” The release includes only ten abridged reasoning summaries for the 372 families. OpenAI says the Riemann-related write-up was edited by humans for readability; it and the Hodge result were exceptions to the standard process.
722 proofs, one question: will any of OpenAI’s AI mathematics actually lead anywhere?
An unreleased, unnamed model produced claimed proofs of results that would each define a career. Sam Altman calls them “claims not yet confirmed by outside mathematicians.” The real question isn’t whether it’s impressive. It’s whether answers nobody understands become discoveries anyone can build on.
Same day: Alon, Bloom, Gowers, Litt, Sawin post a digested, human-verified version. The model for success.
Connes rigidity counterexample challenged within a day — constructed groups fail the required condition. Three rival machine “counterexamples” from different labs now circulate.
~10,000 agents, 88 hours, est. ~$22M at retail. Priority dispute; 25 Fields Medalists sign “A Severe Misalignment” — not saying it’s wrong, saying it’s not understood.
Altman now hedges at announcement — a shift from September. Verification has barely started.
Humans extract the technique, write it up, build on it. This is where downstream discovery comes from.
The question is answered; nobody learns anything reusable. Closes a door without opening a field.
The proof breaks, or proves a statement that doesn’t match the conjecture as mathematicians mean it.
The Unique Games Conjecture is the clearest case. Results like the optimality of Goemans–Williamson for Max-Cut are proved assuming UGC. A correct proof converts them all — no understanding required. A zero-free strip for zeta works the same way for prime-distribution results. Free group factors, Kadison, Mahler would redirect whole programmes — but how depends on the method, which means digestion.
Technology. A Navier–Stokes blow-up proof doesn’t change how anyone designs aircraft; engineering turbulence models never depended on the answer. Near-term consequences are mathematical, not industrial. “AI will cure cancer next” skips several steps.
“Verification abundance, adjudication scarcity” — making proof-checking cheap doesn’t reduce the burden of deciding what’s true and what matters. 722 manuscripts land on a review system built for a trickle, filtered by a selection nobody outside OpenAI made.
Humans re-deriving results, like Alon–Gowers et al. in May
Other people’s work building on these manuscripts
How many unformalized results survive expert checking
Do the Lean statements match the real conjectures?
Do any survive peer review?
Some of it, yes — where a literature is waiting (UGC), a correct proof pays off immediately; where a proof carries a new technique humans digest, it can open a field. Most of it, probably not on its own: at 722 manuscripts with 10 reasoning summaries, the Four Colour pattern is the likely default unless mathematicians are funded and given time. And some will be wrong — OpenAI says so itself. It’s an industry pattern, not one company’s: the forced-Euler result came from an Anthropic researcher, and rival machine-generated Connes “counterexamples” circulate from different labs. The proofs arrived this week. The discoveries, if they come, will arrive at the speed of human understanding.
The significance of the release will depend on more than whether individual statements are true. In mathematics, a proof can matter because it introduces methods other researchers can understand and reuse. If mathematicians can extract such methods from the manuscripts, the work could contribute to new research beyond settling the problems named in the papers.
Some claims could have wider effects if verified. The Unique Games Conjecture, for example, underpins many results in theoretical computer science about the limits of approximation algorithms. A proof could require researchers to revisit conclusions that rely on the conjecture. But that consequence is conditional: the manuscript’s claim first needs independent scrutiny, and the released material does not establish that the broader literature would change.
There is also a less expansive possibility: a result may be correct but offer little reusable insight, or may fail review. The central question is whether mathematicians can verify the work, understand its reasoning and build on it. Publication volume alone does not answer that question.
As an affiliate, we earn on qualifying purchases.
OpenAI’s Earlier Math Releases
This is OpenAI’s fourth major mathematics release this year, according to the source material. In May, the company reported that a model had produced a counterexample to the Erdős unit-distance conjecture. Five mathematicians later published a human-verified, digestible version. That process illustrates one route from AI output to accepted mathematical work: researchers translate and check the result themselves.
An August release called “Ten Advances” included a claimed counterexample to Connes’s rigidity conjecture that was disputed within a day. A critique said the constructed groups did not meet the condition required by the conjecture. The episode shows why a generated proof or counterexample remains a claim until its details are checked.
In September, OpenAI announced a Lean-formalized result concerning finite-time blow-up in the Navier–Stokes equations, produced by what it described as about 10,000 concurrent agents over 88 hours. That work prompted a separate debate: 25 Fields Medalists signed a declaration criticizing the use of famous problems as AI benchmarks without enough attention to human understanding. Their objection, as described in the source material, concerned the aims and practice of mathematical research, not a finding that the proof was wrong.
“A Severe Misalignment of AI in Mathematics.”
— The 25 Fields Medalists who signed the September declaration
mathematical theorem verification tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Independent Checks Still Needed
It is not yet clear which of the 372 result families will withstand independent review, how much of the work can be checked using Lean, or whether the formalizations cover the central claims in each manuscript. OpenAI’s warning about unformalized work and the small number of reasoning summaries mean readers do not yet have a complete, easily assessable account of the model’s reasoning.
The selection process also leaves open how representative the collection is. OpenAI posed roughly 4,000 problems and chose work it judged significant, but the release does not provide an independent assessment of that screening. The model’s identity, training details and broader evaluation process are also not disclosed in the supplied material. No outside confirmation of the headline claims is reported here.
AI-powered mathematical research books
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Mathematicians Must Test the Claims
The next step is for researchers to examine the papers, verify their proofs and identify errors or new techniques. Results with formal Lean proofs may be easier to check mechanically, but formalization does not by itself show that a manuscript addresses the intended conjecture or that its ideas are useful to other researchers.
OpenAI has not, in the source material, announced a timetable for independent evaluations or for releasing the model. The practical test will be whether mathematicians produce verified accounts of particular results and can explain what those results add. Until then, the collection is a large body of AI-generated mathematical claims awaiting evaluation, not a confirmed list of solved problems.
formal proof software for mathematicians
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What did OpenAI release?
It published 722 mathematical manuscripts grouped into 372 families, generated by an unnamed model that has not been released.
Have outside mathematicians verified the results?
Not according to the supplied source material. Altman described them as claims not yet confirmed by outside mathematicians, and OpenAI’s repository cautions that some unformalized results may have issues.
What does Lean formalization mean?
Lean is a proof assistant that can check formalized mathematical arguments against specified definitions and rules. OpenAI says many, but not all, results have Lean formalizations; that does not establish that every manuscript’s claim has been independently accepted.
Could the manuscripts change mathematics?
Potentially, if claims are verified and researchers can understand and reuse the methods. The impact remains unknown: some results may prove significant, some may be correct but yield few new tools, and others may not hold up under review.
Source: ThorstenMeyerAI.com
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
