AI is both the best and worst thing to ever happen to mathematics — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- OpenAI announced its AI agents solved the Navier-Stokes blow-up question in 88 hours after learning two mathematicians had made unpublished progress, confirming the 200-year-old fluid equations do sometimes blow up — a finding that means the equations are imperfect, not that fluids misbehave.
- The blow-up problem sits on the list of seven Millennium Prize Problems, each carrying a $1 million reward; OpenAI separately claims it has made "substantial progress" on another problem from the same list.
- AI proofs tend to produce hard-to-follow, convoluted arguments — unlike the elegant proofs celebrated in Paul Erdős's concept of "The Book" — though a process called formalization lets computers rigorously verify them.
- The article draws a historical parallel to Descartes's 17th-century linking of geometry and algebra, which philosopher Thomas Hobbes dismissed as a "scab of symbols" but ultimately gave rise to calculus and relativity.
- Mathematicians once considered AI an "embarrassment" at math; the article now calls AI's arrival an upheaval comparable to Descartes — producing "many more answers than explanations."
Why it matters: OpenAI demonstrated AI can crack a $1 million Millennium Prize Problem in days — work that resisted mathematicians for centuries. For working mathematicians, the article frames a concrete tradeoff: a tool that can vastly accelerate conjecture-testing while delivering harder-to-interpret proofs.
Ask SkimNews
