🚨 AI may be entering a completely different phase.

OpenAI reportedly began training a new internal model on August 28, which has since solved over 100 long-standing open problems in mathematics, including work on the Navier–Stokes Millennium Prize Problem. Mathematicians like Edward Witten, Timothy Gowers, and Martin Hairer have formed an advisory group to assess these numerous significant results. The rapid pace of new mathematical discoveries by AI raises questions about human evaluation capacity, though these claims await rigorous, independent verification.

Unlike previous AI advancements, this report highlights the formation of an advisory group with prominent mathematicians to handle the sheer volume of new mathematical discoveries, moving the challenge from AI capability to human evaluation capacity.