OpenAI said on 21 September that an internal model it began training on 28 August has resolved more than 100 long-standing open problems across most areas of mathematics, including the Navier–Stokes Millennium Prize problem. No papers or proof artefacts were published alongside the 100-problem claim.
The announcement came with an independent advisory group of nine mathematicians, hosted at the Institute for Advanced Study, which will advise on how such results are assessed, released and attributed. Members include Timothy Gowers, Edward Witten, Martin Hairer, Ravi Vakil and Melanie Matchett Wood.
OpenAI says the group is unpaid, may publish advice the company did not ask for and may criticise it in public, but will not advise on how fast OpenAI pushes its internal mathematics work. The group says it formed after OpenAI approached members about an external board and they chose to set up independently instead, and that it will advise any AI company.
Behind this is a declaration published on 11 September and signed by 27 Fields medallists, among them Terence Tao, Peter Scholze and Pierre Deligne, arguing that racing to solve famous problems as a benchmark damages mathematics. Hairer signed it and now sits on the group.
The 100-problem figure is OpenAI’s own, and for now unverifiable from outside.