OpenAI claims solution to Navier-Stokes Millennium Problem, sparking outcry
The use of 10,000 AI agents to crack a $1 million prize problem raises fears over the devaluation of human genius.
OpenAI has announced that its latest AI system has solved the Navier-Stokes problem, one of the seven Millennium Prize Problems. The breakthrough marks a pivotal moment in the intersection of artificial intelligence and pure mathematics, signaling a shift toward compute-driven discovery.
To reach the solution, which predicts the behavior of fluids and weather, OpenAI deployed approximately 10,000 autonomous agents. The company estimated the compute cost for the operation at roughly $15 million. While the achievement is a technical milestone, the solution relied heavily on existing foundational work by Spanish mathematicians Diego Córdoba and Luis Martínez-Zoroa.
A shift in mathematical discovery
The Millennium Prize Problems were established by the Clay Mathematics Institute in 2000, with each carrying a $1 million reward. Historically, these problems have required decades of human intuition and rigorous peer review. A primary example is Andrew Wiles’s long-term effort to prove Fermat’s Last Theorem. However, the landscape is changing; Google DeepMind has already demonstrated success with International Mathematical Olympiad (IMO) problems, suggesting that AI is moving toward superhuman capabilities in the field.
The erosion of the creative process
The mathematical community has reacted with distress, fearing that "brute-force" compute resources are replacing the creative spark of human research. Critics argue that the ability to spend millions of dollars on compute to solve a problem devalues the intellectual struggle inherent in mathematics. Prof James Robinson of the University of Warwick described the move as "immature playground boasting writ large," noting the potential for significant environmental damage accompanying such massive compute expenditures.
Other academics expressed a sense of professional disorientation. Prof Colva Roney-Dougal of the University of St Andrews described feeling "slightly shell-shocked" by the development. There are growing concerns that the academic incentive structure—centered on publications and undergraduate pedagogy—will collapse if AI can generate proofs faster than humans can understand them. This shift threatens to turn mathematicians into "accountants" whose primary role is to audit AI-generated outputs rather than innovate.
A climate of secrecy
The incident has also highlighted a growing tension regarding intellectual property and collaboration. As AI companies increasingly mine mathematical research to train their models, some researchers are becoming more guarded. Prof Tristan Buckmaster of New York University noted that the current atmosphere in the field is one where "nobody wants to share anything."
What remains to be seen is how the Clay Mathematics Institute will formally validate the AI-generated proof and whether this event will trigger a broader movement to protect human-led research from the encroachment of industrial-scale compute.