OpenAI claims to have solved the Navier-Stokes Millennium Prize problem regarding fluid dynamics by deploying approximately 10,000 agents and tens of millions of dollars in compute over an 88-hour period. The effort put OpenAI in direct competition with NYU professor Tristan Buckmaster and Anthropic researcher Levent Alpöge, who were independently pursuing the proof. The rapid breakthrough has drawn sharp criticism from academic mathematicians, who expressed unease over the company’s aggressive approach to field norms and scientific recognition.
The achievement has also fueled disputes regarding intellectual property and tool usage. Buckmaster questioned whether his previous prompts within OpenAI’s Codex tool influenced the company’s progress. OpenAI spokesperson Laurance Fauconnet categorically denied the claim, stating it was impossible for Buckmaster’s prompts over the preceding two months to have impacted training or model outputs. The incident highlights growing friction between academic researchers and heavily funded AI labs competing for scientific breakthroughs.
Why it matters
AI labs are deploying tens of millions of dollars in compute to solve long-standing fundamental mathematics and hard science problems.
Friction is escalating between academia and AI firms over credit, prompt data usage, and the disruption of traditional scientific research norms.
Source: theverge.com
