OpenAI released findings on 377 mathematical problems spanning algebra, number theory, theoretical computer science, mathematical logic, and topology, following its earlier claim of solving the Navier-Stokes equation, a Millennium Problem with a $1 million prize. The results used an unreleased advanced AI model and were reviewed using the Lean verification system, with average computation taking about three hours per solution. OpenAI said it incorporated feedback from an independent advisory board hosted by the Institute for Advanced Study in Princeton, N.
J., which includes Harvard mathematician Melanie Wood. The board urged transparency, calling for release of AI prompts and chains of thought, and criticized AI companies for testing proprietary models on unsolved problems. Despite this, OpenAI continues internal testing, with research lead Dan Roberts stating it helps build better tools.
NYU mathematician Tristan Buckmaster warned that AI may be completing proofs using mathematicians’ prior work, raising concerns about credit and originality in AI-assisted discovery.
Source: New York Times Top Stories · Summarized by HeadlinesBriefing