🔍 Read the full analysis: What Will OpenAI’s AI Mathematics Amount To 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 manuscripts produced by an unnamed, unreleased model, covering 372 families of mathematical results from roughly 4,000 posed problems. Some claims concern major open problems, but outside mathematicians have not verified the catalogue; the key test is whether its proofs hold up and yield ideas others can use.
OpenAI published 722 mathematical manuscripts on Monday, presenting work by an unnamed, unreleased model across 372 families of related results. The catalogue includes claims involving major open problems, but OpenAI chief executive Sam Altman said the claims have not been confirmed by outside mathematicians—leaving verification and the usefulness of the proofs as the central questions.
According to OpenAI’s post and the project’s GitHub repository, the manuscripts span number theory, geometry, topology, operator algebras, theoretical computer science and mathematical physics. The results came from roughly 4,000 problems posed to the model, with OpenAI selecting work it considered to have an appropriate level of significance. The company says the average result took about three hours of ChatGPT Pro thinking compute. The selection process was conducted by OpenAI, not by independent mathematicians.
The catalogue includes claims about the Unique Games Conjecture, Hilbert’s tenth problem over the rationals, the isomorphism of nonabelian free group factors, the Hodge conjecture for certain abelian varieties, and a zero-free region for the Riemann zeta function to the right of Re(s) = 11/12. If correct, some of these would resolve longstanding problems or establish substantial new results. That importance is conditional: publication by OpenAI is not itself independent verification.
OpenAI released the manuscripts under an Apache-2.0 license. The repository contains Lean formalizations for many, but not all, of the results. Its README cautions that “some of the unformalized results could have issues.” OpenAI also supplied just 10 abridged reasoning summaries for the 372 families. The Riemann zero-free-region write-up and the Hodge result were exceptions to the standard process; the Riemann manuscript 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.
Verification Will Shape the Payoff
The release matters less as a tally of papers than as a test of whether machine-produced mathematics can withstand expert scrutiny and contribute ideas that other researchers can use. A proof can settle a question without giving mathematicians a reusable method. The catalogue’s eventual value will depend on correctness, accessibility and follow-on work, not simply on the number or fame of the claims.
The Unique Games Conjecture illustrates the potential stakes. Many results in theoretical computer science rely on it to establish limits on approximation algorithms, including work on the Max-Cut problem. A valid proof could change how researchers interpret those conditional results. But the manuscript must first be checked carefully, and researchers would still need to understand its reasoning and consequences.
There is also a practical cost to evaluating a large collection. Experts must identify gaps, confirm that formal statements match the problems at issue, and translate unfamiliar reasoning into arguments people can assess. If machine output requires extensive human repair or explanation, the number of manuscripts may overstate how much usable mathematics has been produced. The central outcome is therefore not whether AI can generate claims, but whether mathematicians can verify and build on them.
mathematics problem solving software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Earlier Releases Offer Caution
This is OpenAI’s fourth major mathematics release of the year, according to the source material, and earlier episodes show why claims need to be examined individually. In May, an OpenAI 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 the same day. That process turned machine output into work the mathematical community could evaluate.
OpenAI’s August collection, called “Ten Advances,” included a claimed counterexample to Connes’s rigidity conjecture. A critique published within a day argued that the constructed groups did not meet the conditions required by the conjecture. The dispute is a reminder that a plausible-looking proof or counterexample can fail because it addresses a subtly different statement.
In September, OpenAI announced a Lean-formalized proof concerning finite-time blow-up in the Navier–Stokes equations, produced, the company said, by about 10,000 concurrent agents over 88 hours. That announcement came amid a priority dispute involving separate work on forced Euler equations by Levent Alpöge and Tristan Buckmaster. Three days later, 25 Fields Medalists signed a declaration titled “A Severe Misalignment of AI in Mathematics.” Their stated concern was not that the announced proof was wrong, but that treating famous problems as benchmarks without human understanding could conflict with the aims of mathematics.
As an affiliate, we earn on qualifying purchases.
Which Manuscripts Will Hold Up
Independent verification remains outstanding for the catalogue’s claims, and the source material does not give a result-by-result assessment by outside experts. It is not yet clear how many manuscripts will prove correct, whether any formalizations cover the hardest claims in full, or how much revision may be needed. A Lean formalization can help check a precisely stated argument, but it does not by itself establish that the formalized statement is the intended conjecture or that the result yields broader insight.
The selection criteria and the complete path from roughly 4,000 proposed problems to 372 families are also not independently documented in the provided material. OpenAI says it selected results for significance, but outsiders did not make that selection. Nor do the 10 abridged summaries provide a full guide to all 372 families. It remains unclear which claims will attract sustained expert review, and whether researchers will be able to extract reusable methods from the proofs.
AI-assisted mathematical research books
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Outside Review Is the Next Test
The next meaningful milestone is independent mathematical review: experts checking whether each argument is valid, whether it proves the stated result, and whether its methods can be understood and reused. Some claims may be confirmed, revised or rejected; others may remain difficult to assess for an extended period. The repository’s publication makes the manuscripts available for scrutiny, but it does not set a timetable for that work.
For readers, the catalogue should be treated as a collection of research claims rather than a list of settled breakthroughs. The May Erdős example shows one possible path: mathematicians digest the output, verify it and present a form the field can evaluate. Whether any of the new manuscripts follow that path—and whether they lead to further results—will become clearer only as outside researchers engage with them.
advanced mathematical logic textbooks
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What did OpenAI release?
OpenAI published 722 mathematical manuscripts grouped into 372 families of related results. The company says they were generated by an unnamed, unreleased model from roughly 4,000 problems.
Have mathematicians verified the major claims?
Not according to the source material. Sam Altman described the results as claims not yet confirmed by outside mathematicians. OpenAI’s repository also warns that some unformalized results could have issues.
What are some of the problems addressed?
The manuscripts include claims concerning the Unique Games Conjecture, Hilbert’s tenth problem over the rationals, the Hodge conjecture for certain abelian varieties, and a zero-free region for the Riemann zeta function. Their correctness has not been established in the supplied material.
Why does a proof need more than verification?
A proof may settle a statement but still offer little that researchers can reuse. Its broader mathematical value depends on whether people can understand its methods and use them to develop further results.
What should happen next?
Outside mathematicians need to examine the arguments, check that they prove the stated claims and assess whether the reasoning is useful. The release provides material for that review, but no timeline or independent verdict is given.
Source: ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
