OpenAI Claims AI Solution to Millennium Prize Problem

OpenAI announced on Tuesday that one of its internal AI models has solved the Navier-Stokes problem, one of seven Millennium Prize Problems established by the Clay Mathematics Institute in 2000. The problem, which dates back 200 years, concerns the motion of fluids and whether the equations describing them fail under certain conditions. OpenAI stated that the solution was reached in 88 hours using up to 10,000 AI agents. The company published a Lean proof to allow independent experts to verify the results. The Clay Mathematics Institute offers a 1 million dollar prize for the first correct solution to each of the seven problems, only one of which had been solved previously.

The announcement has sparked a dispute regarding the origin of the discovery. New York University professor Tristan Buckmaster and Anthropic researcher Levent Alpöge had been working on the same problem and had recently published progress. Buckmaster suggested that OpenAI acted opaquely and only began its focused effort after hearing rumors of their work. OpenAI denied using their specific work or prompts, though it admitted it cannot rule out that de-identified data from product usage helped improve its models. OpenAI also noted that its work on the problem began on September 1 after rumors of a potential resolution surfaced.

Some mathematicians have expressed concern over the impact of AI on the field. Terence Tao, a professor at UCLA, compared the use of AI to a machine lifting weights at a gym, arguing that it removes the instructive struggle necessary for human understanding. He suggested that the current environment is becoming a frenetic competition based on production quotas rather than mathematical value. While center-leaning reports focus on the tension between AI progress and human understanding, center-left reports emphasize the potential for loss of control over AI and the high energy costs associated with such computations.