Mathematicians learn little from AI completing unsolved problem : NPR

Key Points:

• OpenAI announced that its AI solved the Navier-Stokes problem, one of the Millennium Prize Problems, but mathematicians find the 166-page AI-generated proof difficult to understand and not yet helpful for advancing human knowledge.
• The Navier-Stokes problem concerns fundamental equations describing fluid flow, and solving it could lead to improved models in physics and engineering, such as turbulence and aircraft lift.
• OpenAI deployed around 10,000 AI agents for 88 hours, using roughly 300 billion tokens at an estimated cost of $15-20 million, to generate the solution, sparking controversy over alleged knowledge of ongoing human research and rushed publication.
• While the AI-generated proof is formally verified through Lean code, mathematicians criticize the lack of clarity and human-readable explanation, emphasizing that mathematical progress requires more than just a correct answer.
• Despite the criticism, experts acknowledge the growing role of AI in mathematics but stress the importance of collaboration and clear communication between AI systems and human researchers for meaningful advancements.