started · updated
OpenAI releases hundreds of AI-generated mathematical manuscripts
OpenAI has released a massive repository of 719 to 722 mathematical manuscripts generated by an unreleased, internal frontier AI model. The collection is organized into 372 families of results covering diverse fields such as algebraic geometry, number theory, and theoretical computer science.
Among the claimed breakthroughs are advancements related to the Riemann hypothesis, specifically a demonstration of the 'quasi-Riemann hypothesis', and the Birch–Swinnerton-Dyer conjecture. While the scale of the release is unprecedented, it has sparked significant controversy within the scientific community. Many mathematicians have expressed concern over the lack of transparency, as the underlying model remains proprietary and inaccessible for independent verification.
Critics, including the Association for Human Mathematics and prominent mathematicians like Terence Tao, have criticized the method of release. They argue that the simultaneous dumping of hundreds of unverified papers bypasses traditional peer-review processes and threatens the collaborative nature of mathematical research. Some researchers have described the event as a 'mathocalypse' or a 'demonstration of power' rather than a scientific contribution.
OpenAI noted that approximately 42% of the top-line results have been formalized using the Lean programming language to allow for computer-assisted logical verification. However, the company has already had to retract or revise several manuscripts due to errors, such as sign mistakes in algebraic geometry proofs.
Entities
Alex Kontorovich · Association for Human Mathematics · Clay Mathematics Institute · GitHub · Institute for Advanced Study · Martin Bridson · OpenAI · Princeton · Sam Altman · Terence Tao
Claims
What the coverage asserts, and how many sources carry each claim.
- [● 4 SOURCES] OpenAI retracted three manuscripts due to a sign error that invalidated an argument. nacion.com.uy · mattbaker.blog · www.tokenpost.com · fakti.bg
- [● 3 SOURCES] None of the released proofs fully solve any of the five Millennium Prize Problems. nacion.com.uy · eveningreport.nz · fakti.bg
- [● 4 SOURCES] OpenAI is working with an advisory council of mathematicians hosted at the Institute for Advanced Study in Princeton. vanguardia.com.mx · nacion.com.uy · www.tokenpost.com · tech.everyeye.it
- [● 7 SOURCES] The manuscripts cover 372 families of results in fields such as number theory and algebraic geometry. nacion.com.uy · eveningreport.nz · diariosanrafael.com.ar · www.tokenpost.com · fakti.bg · +2 more
- [● 7 SOURCES] Approximately 42% of the top-line results have been formalized using the Lean programming language. mattbaker.blog · nacion.com.uy · www.tokenpost.com · fakti.bg · tech.everyeye.it · +2 more
- [● 2 SOURCES] The release includes a demonstration of the 'quasi-Riemann hypothesis'. nacion.com.uy · eveningreport.nz
- [● 5 SOURCES] Each result required an average of three hours of reasoning from ChatGPT Pro. nacion.com.uy · diariosanrafael.com.ar · www.tokenpost.com · fakti.bg · tech.everyeye.it
- [● 9 SOURCES] OpenAI released a repository containing 722 mathematical manuscripts. nacion.com.uy · eveningreport.nz · diariosanrafael.com.ar · www.tokenpost.com · fakti.bg · +4 more
- [○ 1 SOURCE] Sam Altman stated that the world should accept certain negative outcomes of AI to realize the technology's benefits.
- [○ 1 SOURCE] Scientists express concern that autonomous AI systems could publish discoveries before human researchers, impacting academic credit. www.hardware.com.br
- [○ 1 SOURCE] Nearly all the reported mathematical results were produced by a single AI agent using only one prompt. www.tivi.fi
- [○ 1 SOURCE] The reported advancements include the four-dimensional Kakeya conjecture and progress toward the Riemann hypothesis. www.tivi.fi