AI claimed a solution to the 167-year-old Riemann Hypothesis — and a human did, too
AI claimed a solution to the 167-year-old Riemann Hypothesis — and a human did, too
Anthropic mathematician Levent Alpöge announced a Lean-compilable solution to the Riemann Hypothesis, which dates to 1859 and is a Millennium Prize problem, saying he worked through the argument with Claude; he and NYU's Tristan Buckmaster also released a proof related to the Navier-Stokes equations. Separately, Stanford professor and American Mathematics Association president Ravi Vakil posted a General Riemann Hypothesis proof by Stanford postdoc Alessandro Masullo, who said it is the product of twelve years of work and "all human, with no AI." Alpöge acknowledged releasing his result without a human-intelligible explanation and committed to pursuing one.
It captures a same-week collision between an AI-assisted proof claim and a fully human one, a useful snapshot of the current race between frontier labs and mathematicians.
Source: TechCrunch · AI · techcrunch.com