🔍 Read the full analysis: Can 722 Proofs Point OpenAI’s AI Mathematics Toward A Bigger Goal? on ThorstenMeyerAI.com
Get hardware and tech essentials 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 attributed to an unnamed, unreleased model, covering 372 families of results drawn from about 4,000 problems. The release includes claims about longstanding conjectures, but OpenAI says outside mathematicians have not confirmed them; whether the work yields usable new methods remains unknown.
OpenAI published 722 mathematical manuscripts on Monday, presenting work by an unnamed, unreleased model across 372 families of related results. The papers include claims about several longstanding problems, but OpenAI chief executive Sam Altman said the results are not yet confirmed by outside mathematicians, leaving their validity and potential impact unsettled.
OpenAI’s release and GitHub repository say the manuscripts span fields including number theory, geometry, operator algebras, topology, theoretical computer science and mathematical physics. The results were selected from roughly 4,000 problems posed to the model. OpenAI filtered the output for what it described as an appropriate level of significance; the selection was made inside the company, not by an independent mathematical panel. The reported average was about three hours of ChatGPT Pro thinking compute per result.
The catalogue includes claims of a proof of the Unique Games Conjecture, a resolution of Hilbert’s tenth problem over the rationals, and a proof that all nonabelian free group factors are isomorphic. Other listed results concern a zero-free region for the Riemann zeta function to the right of Re(s) = 11/12, the Hodge conjecture for CM abelian varieties, and conjectures in convex geometry. These are claims in the released material, not established solutions.
OpenAI says many, but not all, of the results have Lean formalizations, which can help check proof steps using a computer proof assistant. Its repository warns that “some of the unformalized results could have issues.” The release also contains ten abridged reasoning summaries for 372 families. OpenAI said two manuscripts—the Riemann zero-free-region work and the Hodge result—did not follow the standard process; the Riemann write-up was edited by humans for readability.
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.
From Machine Proofs to New Mathematics
The central question is not just whether a model can produce a correct proof. In mathematics, a result often matters most when its argument supplies ideas and techniques other researchers can reuse. A proof that settles a famous question but remains difficult to understand may have less influence on the field than a shorter argument that opens new lines of work.
The Unique Games Conjecture illustrates the potential reach if the claim survives scrutiny. Many theoretical computer science results use the conjecture to establish limits on approximation algorithms. A verified resolution could prompt researchers to revisit conclusions that depend on it. But that consequence is conditional: the released claim has not been independently established, and its significance cannot be assessed from the headline result alone.
The practical test is whether mathematicians can check, explain and build on the work. OpenAI’s earlier mathematics releases have produced different outcomes, including a result that researchers described as human-verified and another claim that was challenged over whether its construction met the conjecture’s conditions. The current collection’s size does not itself show that it will produce broader discoveries.
mathematical proof assistant software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
OpenAI’s Earlier Math Results
This is OpenAI’s fourth major mathematics release this year, following releases in May, August and September, according to the source material. In May, the company’s model produced a counterexample to the Erdős unit-distance conjecture. Five mathematicians—Noga Alon, Thomas Bloom, Tim Gowers, Daniel Litt and Will Sawin—posted a human-verified account of the result. That process offers one example of how machine-generated work can become useful: researchers translate the output into a form they can evaluate and then check it.
An August release, called “Ten Advances,” had a more contested result. A claimed counterexample to Connes’s rigidity conjecture was challenged within a day, with critics arguing that the constructed groups did not satisfy a condition required by the conjecture. In September, OpenAI announced a Lean-formalized Navier–Stokes blow-up proof generated by about 10,000 concurrent agents over 88 hours. The announcement prompted a dispute about research priorities: 25 Fields Medalists signed a declaration criticizing the use of famous problems as AI benchmarks without adequate human understanding. The signatories’ criticism concerned the approach to AI mathematics, not a finding that the proof was wrong.
formal verification tools for mathematicians
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Independent Checks Still Pending
The release does not establish that any of its most prominent claims is correct. Independent review of the 722 manuscripts is not described as complete, and the material provided does not identify which results have been checked by mathematicians unaffiliated with OpenAI. Lean formalizations are available for many results, but not all; formalization also does not, by itself, establish that a paper’s stated theorem is the result researchers intended to prove.
It is also unclear how the company defined significance when choosing results from roughly 4,000 problems, how much of each manuscript can be independently reproduced, and whether the ten abridged summaries give enough detail for experts to assess the full set. The source material does not provide a timeline for external review or say which research groups are undertaking it. The status of the claimed major results should therefore remain provisional.
AI-powered theorem proving software
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 independent mathematical scrutiny: checking the arguments, testing any formalized proofs, and determining whether the stated conclusions match the original problems. For claims that withstand review, researchers will also need to extract the reasoning and decide whether it offers methods that can support further work. The May Erdős result suggests one possible route, with mathematicians producing a checked and readable account of machine output.
No independent verdict or review schedule for the new catalogue was included in the supplied material. Until those checks are reported, the 722 manuscripts are best treated as a set of potentially significant claims rather than a confirmed collection of discoveries. Their longer-term value will depend on both correctness and whether other mathematicians can understand and use the proofs.
As an affiliate, we earn on qualifying purchases.
Key Questions
What did OpenAI release?
OpenAI published 722 mathematical manuscripts in 372 families, generated by an unnamed model that has not been released. The work was selected from roughly 4,000 problems posed to the model.
Have the claimed results been verified?
Not as a collection. Sam Altman described them as claims that outside mathematicians have not yet confirmed. OpenAI also cautioned that some results without formalizations could have issues.
What is the most consequential claim?
The catalogue includes a claimed proof of the Unique Games Conjecture, which underpins many results about the limits of approximation algorithms. Its consequences depend on independent experts verifying the proof.
Does a computer-checkable proof settle whether a result matters?
No. A Lean formalization can help check proof steps, but mathematicians still need to assess whether the argument establishes the intended claim and whether its methods can be understood and reused.
What happens next?
Researchers need to examine the manuscripts, verify the arguments and identify any reusable techniques. The supplied material gives no timetable for external review or independent verdicts.
Source: ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
