The lesson from AlphaFold for mathematicians

The reaction to OpenAI’s claimed Navier–Stokes solution reminds me of AlphaFold. Protein structure prediction used to be a major research specialty. For students it was often one PhD to get one protein.Then AlphaFold made high-quality predictions available to basically everyone for free. Protein scientists didn’t disappear. The bottleneck just moved from predicting the structure to understanding what it means. If AI starts solving frontier math problems, maybe the same thing happens to mathematics. Human value may shift toward judgment, verification, interpretation, and choosing problems... but AI may automate those too. The field could simply need fewer mathematicians for the same amount of output.