Medicine's AI misalignment problem, through the lens of the Navier-Stokes debacle

davidjhwu · x · 2026-09-17

The author examines the Navier-Stokes Millennium Prize controversy: after learning a mathematician using Codex was close to a solution, OpenAI reportedly deployed 10,000 agents and millions of dollars of compute to brute-force a possible answer within days. The researcher alleged foul play; OpenAI said it didn't explicitly train on his chats but "cannot rule out" that de-identified data from his usage helped improve its models. The piece extends this to medicine, arguing the field faces its own AI misalignment problem around IP boundaries, data ethics, and the displacement of domain experts.

Original post →

More from AGI Musings

AGI Musings channel →