OpenAI turns to independent mathematicians after AI solves over 100 open problems

OpenAI says a new internal model has resolved more than 100 long standing open problems across mathematics, prompting the company to work with an independent group of mathematicians on how the results should be reviewed and released.

The model began training on August 28 and was also responsible for OpenAI’s recently published solution to the Navier Stokes Millennium Prize problem. 

OpenAI has described the system as significantly more capable than GPT-6 Astra and said its progress in mathematics has surprised researchers inside the company.

OpenAI has not yet publicly detailed or released the more than 100 additional claimed solutions, meaning their significance and validity cannot yet be independently assessed.

The company is now working with the newly formed Advisory Group on Mathematics and Artificial Intelligence, an independent group hosted at the Institute for Advanced Study. 

Its members include Timothy Gowers, Martin Hairer, Ravi Vakil, Edward Witten and Melanie Matchett Wood.

The group said its immediate task is advising OpenAI on how to coordinate the release of what the company reports is a large number of significant mathematical results. 

It will also advise AI companies on the responsible presentation of mathematical research and publish its recommendations publicly.

The group operates independently from OpenAI and its members are not paid by the company. It has no decision making authority over OpenAI and will not advise the company on how quickly to advance its internal mathematics capabilities.

The move follows growing debate among mathematicians over the use of unresolved research problems to benchmark increasingly capable AI systems. 

OpenAI said it wants mathematicians involved in assessing the significance of new results and determining how they should be communicated as its systems become more capable of producing original mathematical research.