In recent months, unreleased frontier AI systems have fully formalised Andrew Wiles's proof of Fermat's Last Theorem, solved the Navier-Stokes problem and on October 26 felled another Clay Millennium Prize problem.
Mathematics is in a collective vertigo: AI keeps solving Millennium Prize problems, and one PhD student wrote that he had been screaming internally for days.
A Fields Medal winner closed a talk by saying we need to develop new theories of care and love. Coming from that mouth, it says more about the panic than any headline.
The author's way out is to swap the checklist model of progress for a generative encounter model: stop ticking off famous problems and go back to defining problems through real-world encounters. That shift feels more worth chewing on than the question of whether AI replaces mathematicians.
Curated from high-quality sources, with concise summaries and key takeaways.
