OpenAI-affiliated accounts reported that an AI-assisted effort produced a result related to the Navier-Stokes existence and regularity problem, one of the Millennium Prize Problems. The claim centers on a collaborative system involving approximately 10,000 agents trained over a year using multi-agent reinforcement learning.
- Ethan Knight stated the solution resulted from ~10,000 agents working together, utilizing massive unstructured parallel test-time compute with models deciding how to organize themselves.
- The training approach involved multi-agent reinforcement learning developed over roughly one year.
- Public interpretation framed the announcement as a potential AI-generated proof regarding finite-time singularity or blow-up in fluid dynamics.
- Acceptance by the mathematical community remains unresolved, and no formal proof sketch, theorem statement, or independent verification was provided in the source tweets.
- The "88 hours" detail cited in some discussions originated from satirical posts rather than the primary OpenAI-linked statements.
The announcement has triggered significant debate regarding the role of large-scale parallel inference compute in frontier research and whether such agentic systems can contribute to serious mathematical problems.