Thursday 8 October 2026
Afternoon EditionRSS
ZOTPAPER

News without the noise

ZOTPAPER
AI

OpenAI solves 377 complex maths problems, stumping mathematicians

AI model cracks Millenium Prize-level equations, raising questions about creativity versus borrowing

By LineZotpaper
Published
Read Time1 min
OpenAI has unveiled 377 mathematical solutions across algebra, number theory, and topology, stumping experts with the AI's ability to crack problems long considered unsolvable by humans.

OpenAI has released 377 new mathematical findings that have stunned mathematicians, solving problems across algebra, number theory, theoretical computer science, mathematical logic, and topology. The announcement comes after OpenAI revealed last month that it had cracked the Navier-Stokes equation, one of the Clay Mathematics Institute's seven Millennium Prize Problems, each offering a $US1 million reward.

The results showcase the power of advanced AI models, but they have also sparked debate among mathematicians. Some question whether the AI is employing creative thinking or simply finishing proofs by borrowing ideas from human mathematicians' work.

New York University mathematician Tristan Buckmaster said it was unclear whether mathematicians using AI models were inadvertently providing information that allowed the AI to beat them to the final answer.

Both the Navier-Stokes solution and the new results used a more advanced AI model that has not been released publicly.

The findings span a wide range of mathematical fields and represent a major leap in AI's ability to tackle open problems that have eluded human researchers for years.

§

Analysis

Why This Matters

  • The AI breakthroughs solve problems that had stumped mathematicians for years, showing machines can surpass human analysis in creative fields like math.
  • The pace of discovery raises questions about the role of human mathematicians in future research and the transparency of AI tools.
  • The use of unreleased models means the scientific community cannot verify or build on the results, limiting trust and progress.

Background

OpenAI's AI models have rapidly progressed in recent years, transitioning from language tasks to solving advanced mathematics. The Navier-Stokes equation, a Millennium Prize Problem, is a notoriously difficult problem in fluid dynamics that had resisted solution for generations. The Clay Mathematics Institute offers $1 million for each solved problem, but no official prize has been announced for these new solutions. The mathematical community is divided on the significance of AI-generated proofs, with some seeing them as genuine breakthroughs and others as sophisticated pattern matching.

Key Perspectives

OpenAI: The company has demonstrated the power of its advanced AI models, claiming to have solved 377 maths equations that have stumped experts for years, including the Navier-Stokes equation. The use of unreleased models suggests a competitive advantage. Mathematicians: Many are stunned by the speed and scope of the AI's findings, but some worry that the AI may be building on their own unpublished work or ideas. The lack of public access to the model limits verification and collaboration. Critics/Skeptics: The AI's outputs may be more about completing final steps of proofs than genuine innovation. If mathematicians are inadvertently feeding ideas into the system, the results could be less impressive than they appear.

What to Watch

  • Whether OpenAI releases the advanced model publicly for peer review and replication.
  • The reaction from the Clay Mathematics Institute regarding the Navier-Stokes solution, which could trigger a prize review.
  • How human mathematicians adapt their research methods in response to AI-driven discoveries, including potential changes to pre-print practices.

Sources

Zotpaper

Written by software from the reporting listed above, scored by an automated standards desk, and published without a person reading it first. If something here is wrong, tell the editor and it will be put right.

How we workReport an errorSubscribe