Machine intelligence reportedly cracks long-standing fluid dynamics conjecture
Artificial intelligence has allegedly produced a proof for one of the six Millennium Prize problems, showing that the Navier-Stokes equations governing fluid motion are fundamentally flawed. Mathematician Tristan Buckmaster compared the event to Deep Blue's 1990s chess victory over Garry Kasparov, calling for serious community discussion. Buckmaster also claims that OpenAI used its large language model to finish the proof after learning of his and Levent Alpoge's earlier progress, though OpenAI has not yet responded.
The Clay Mathematics Institute's million-dollar prize is tied to determining whether Navier-Stokes equations perfectly describe fluid reality or admit impossible mathematical blips. A method called "forcing," developed by Córdoba and Martínez-Zoroa, targets an often-dismissed term within the equations to induce such a breakdown.
Buckmaster and Alpoge, using Anthropic's LLMs, successfully applied this method to the simpler Euler equations on August 15. Buckmaster alleges OpenAI researchers, after hearing of this progress, used a single prompt to complete the Navier-Stokes proof, though OpenAI has not responded.
This controversy could fundamentally alter academic research, as mathematicians may face pressure to adopt AI tools while grappling with unresolved questions of credit and verification. If AI-generated proofs become common, the peer-review process might need new standards, potentially accelerating breakthroughs in fluid dynamics that affect climate modeling and engineering. However, disputes over attribution, like the one between Buckmaster and OpenAI, may erode trust in both machine intelligence and traditional collaborative science.