🔍 Read the full analysis: OpenAI’s AI Mathematics Poses A Question: What Comes After 722 Proofs? on ThorstenMeyerAI.com
Get tech for your team 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 published 722 mathematical manuscripts generated by an unnamed, unreleased model, covering 372 families of results drawn from about 4,000 problems. The collection includes claims about major open problems, but external mathematicians have not yet confirmed them, and OpenAI cautions that some unformalized results may contain errors. The longer-term question is whether researchers can verify and use the work, not just whether the manuscripts produce answers.
OpenAI on Monday published 722 mathematical manuscripts generated by an unnamed model it has not released, presenting claims across 372 families of results drawn from roughly 4,000 problems. The collection includes purported solutions or advances on several famous open questions, but the claims have not been confirmed by outside mathematicians, and the company warns that some results lack formal verification.
The manuscripts span number theory, geometry, operator algebras, topology, theoretical computer science and mathematical physics. OpenAI says the results were selected from about 4,000 problems for what it judged to be an appropriate level of significance. That selection was made inside the company; the published collection does not establish that independent mathematicians reviewed the full set before release. OpenAI published the work under an Apache-2.0 license.
Among the manuscripts are claims concerning the Unique Games Conjecture, Hilbert’s tenth problem over the rationals, the isomorphism of nonabelian free group factors, and a zero-free region for the Riemann zeta function to the right of Re(s) = 11/12. The collection also includes a result about the Hodge conjecture for CM abelian varieties and work on Mahler conjectures in convex geometry. These are claims in the manuscripts, not independently established breakthroughs.
OpenAI reports that many results have Lean formalizations, but not all do. Its repository README cautions that “some of the unformalized results could have issues.” The release provides ten abridged reasoning summaries for 372 families. The Riemann write-up was edited by humans for readability, according to the source material; the Riemann and Hodge results are described as exceptions to the usual procedure. The average result used about three hours of ChatGPT Pro thinking compute, though that figure does not establish the total effort or the reliability of any proof.
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.
Verification Will Determine the Value
The release puts the emphasis on a distinction familiar to mathematicians: a correct proof is not automatically a useful discovery. A proof can settle a question while leaving little that other researchers can apply. Its lasting effect often depends on whether mathematicians can identify, understand and reuse the ideas behind it.
The source material points to OpenAI’s May work on the Erdős unit-distance conjecture as a more constructive pattern: mathematicians produced a digested version of the model’s output and checked it. By contrast, an August claim involving Connes’s rigidity conjecture was challenged within a day, with a critique arguing that the constructed groups did not meet the conjecture’s required condition. Those examples show why the new catalogue must be evaluated result by result, rather than treated as a single verdict on AI mathematics.
If verified, some results could affect research well beyond the original questions. The Unique Games Conjecture, for example, underpins a substantial body of theoretical computer science about the limits of approximation algorithms. But the consequences depend on whether the manuscript proves the precise statement mathematicians study and whether its reasoning can be checked and used. Independent scrutiny, not the size of the catalogue, will determine its mathematical importance.
mathematics problem solver software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Earlier Releases Set a Cautious Precedent
This is described in the source material as OpenAI’s fourth major mathematics release this year. Its previous announcements offer both a positive example and a warning. In May, the model produced a counterexample to the Erdős unit-distance conjecture, which had been proposed in 1946. Five mathematicians—Noga Alon, Thomas Bloom, Tim Gowers, Daniel Litt and Will Sawin—posted a human-verified account that made the result easier to evaluate.
OpenAI’s August “Ten Advances” release drew a faster challenge. A claimed counterexample to Connes’s rigidity conjecture was disputed after critics said the groups in the construction did not satisfy a condition required by the conjecture. The source also describes a September announcement of a Lean-formalized Navier–Stokes blow-up proof, generated with about 10,000 concurrent agents over 88 hours. That claim arrived amid a priority dispute with separate work on forced Euler equations by Levent Alpöge and Tristan Buckmaster.
After the Navier–Stokes announcement, 25 Fields Medalists signed a declaration titled “A Severe Misalignment of AI in Mathematics,” according to the source. Their stated concern was about the purpose and practice of using famous problems as benchmarks without human understanding—not a finding that the proof was wrong. That debate frames the current release: the field is weighing both correctness and what mathematical work is for.
““A Severe Misalignment of AI in Mathematics.””
— The 25 Fields Medalists who signed “A Severe Misalignment of AI in Mathematics”
As an affiliate, we earn on qualifying purchases.
Which Manuscripts Will Survive Review
No outside verification of the full collection is established in the supplied material. It does not say which mathematicians have checked individual manuscripts, whether the Lean formalizations cover each claimed result completely, or how many manuscripts may be revised or withdrawn after scrutiny. The ten abridged summaries also represent only a small portion of the 372 result families, limiting what readers can infer from the summaries alone.
It remains unclear whether the most striking claims prove the exact conjectures and formulations that researchers care about, and whether any proofs contain reusable methods. The model’s identity, technical design and full problem-selection process have not been disclosed in the source material. Until independent researchers examine the work, the number of manuscripts cannot be treated as a count of verified discoveries.
As an affiliate, we earn on qualifying purchases.
Mathematicians Must Test the Claims
The next step is independent checking of individual manuscripts, including their statements, proof details and formalizations where available. Researchers will need to determine which claims are correct, which require repair, and whether any contain ideas that can support further work. The source material does not give a timetable for those reviews or identify a formal process for coordinating them.
As that work proceeds, the most informative developments will be specific: a result independently verified, a counterexample or gap identified, or a proof rewritten into a form other mathematicians can build on. OpenAI’s catalogue establishes that the company has released a large body of mathematical output. What comes after the 722 manuscripts depends on verification and human interpretation, and remains unresolved.
advanced math textbooks for researchers
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What did OpenAI publish?
OpenAI published 722 mathematical manuscripts grouped into 372 families of results, generated by an unnamed model from roughly 4,000 problems.
Have mathematicians verified the results?
The supplied source material does not establish outside verification of the collection. OpenAI says some results are unformalized and warns that they could have issues, so the claims need to be checked individually.
What are some of the most prominent claims?
The manuscripts include claims involving the Unique Games Conjecture, Hilbert’s tenth problem over the rationals, free group factors, the Riemann zeta function and the Hodge conjecture for CM abelian varieties. These remain manuscript claims unless and until they are independently confirmed.
Why does a proof need to be useful as well as correct?
A proof may settle a question without giving researchers methods they can reuse. Its broader impact depends on whether mathematicians can understand the reasoning and apply its ideas to other problems.
What happens next?
Mathematicians will need to examine the manuscripts and formalizations, confirm or challenge the claims, and assess whether the work offers reusable ideas. No review timetable is specified in the supplied material.
Source: ThorstenMeyerAI.com
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
