OpenAI announced plans to release hundreds of unsolved math problems on GitHub, marking the company's latest effort to share AI-generated results. It is the company's first major update in mathematics since August, when it convened 40 mathematicians to discuss AI's potential in the field.
The release includes solutions to over 100 long-standing open problems across mathematics, according to Lindsay McCallum, an OpenAI spokesperson. This compares with the earlier release of 10 problems in August, which was shared through a blog post.
The solutions are generated by OpenAI's new internal model, which was trained to resolve complex mathematical challenges, including the Navier-Stokes Millennium Prize problem. Availability of the GitHub release is set for Tuesday, though a specific date has not been announced.
"On August 28, we began training a new internal model. In addition to resolving the Navier-Stokes Millennium Prize problem, this model has now resolved more than 100 long-standing open problems across most areas of mathematics," said McCallum.
The company is working to responsibly release the next math results, drawing on advice from the Institute for Advanced Study's Advisory Group on Mathematics and Artificial Intelligence.
The announcement follows a series of controversies over previous releases, including accusations that OpenAI front-ran a solution to a Millennium Prize problem. Mathematicians feel the field has become a playground for AI companies like OpenAI and Anthropic, with traditional scientific processes for attribution and verification being sidelined.
OpenAI did not say how it will ensure proper attribution for the results, and mathematicians remain concerned about the lack of transparency. The company has not set a release time, and many feel it is not aligned with the Leiden declaration, a call for AI companies to meet mathematicians’ standards.
Source: wired
