Updated
Updated · WIRED · Oct 6
OpenAI Plans 100-Plus Math Solution Dump on GitHub, Angering Researchers
Updated
Updated · WIRED · Oct 6

OpenAI Plans 100-Plus Math Solution Dump on GitHub, Angering Researchers

3 articles · Updated · WIRED · Oct 6

Summary

  • Hundreds of AI-generated solutions to unsolved math problems could be posted on GitHub as soon as Tuesday, according to people familiar with OpenAI’s plans, despite mathematicians urging the company to publish papers explaining the work.
  • OpenAI says a model it began training on August 28 has resolved the Navier-Stokes Millennium Prize problem and more than 100 long-standing open problems, but spokesperson Lindsay McCallum said no release time has been set.
  • Around 40 mathematicians met OpenAI in August and warned against releasing results through blogs, tweets or other piecemeal posts, saying that format makes claims harder to verify, absorb and properly credit.
  • September disputes deepened the backlash: NYU mathematician Tristan Buckmaster accused OpenAI of front-running work tied to Anthropic, reinforcing complaints that AI labs are sidelining academic norms in a race to showcase capabilities.
  • The clash has pushed mathematicians toward new repositories such as Hexagon and Palomar and sharpened fears that a few AI companies are concentrating too much power over how mathematical results are produced and released.

Insights

Why is OpenAI hiding its monumental Navier-Stokes breakthrough behind cryptic blog posts instead of facing rigorous academic peer review?
If ten thousand AI agents solve a Millennium Prize math problem, who truly deserves the credit and the million-dollar reward?
Could an avalanche of AI-generated proofs completely break the traditional scientific review process before human experts can verify the truth?