Mathematicians say OpenAI doesn't understand its own Navier-Stokes proof
decoder ·
Mathematicians have criticized OpenAI's recent claim of solving the Navier-Stokes problem, arguing that the proposed proof lacks substantive depth despite its formal verification in Lean. Experts note that while the approach superficially resembles work by Spanish mathematicians, detailed analysis reveals significant divergences in methodology. The controversy is further complicated by accusations from NYU’s Tristan Buckmaster, who alleged OpenAI utilized prior unpublished work, a claim the company denies. Additionally, OpenAI released approximately 400 additional results on GitHub, a move that frustrated many in the mathematical community for appearing to trivialize rigorous academic standards. A major point of contention involves the role of AI agents, with researcher Noam Brown admitting that it is unclear how much the massive swarm of 10,000 instances actually contributed. Brown suggests that less than ten percent of the success stems from the multi-agent setup, attributing most gains to the base model's endurance and parallel processing capabilities. Consequently, critics argue that OpenAI does not fully comprehend the mechanics or value of its own computational efforts. The Clay Mathematics Institute continues to review the potential solution as the debate over AI's role in high-level mathematics intensifies.
decoder