What a 166-page AI-authored proof of fluid blow-up teaches the AI public about the difference between solving a problem and understanding the world it lives in.
For two centuries, nobody could prove whether smooth fluid flow can suddenly break down — a Clay Mathematics Institute Millennium Prize problem worth one million dollars. I argued the question itself was malformed. Then, in September 2026, an AI answered it anyway, inside the exact framework I’d called broken. This essay is the honest reckoning that followed: what the AI actually built, what it does and doesn’t prove, and the real distinction between a mathematical claim about an idealization and a physical claim about the world.