Tao expected the Riemann Hypothesis to fall by 2050 — it took AI only 35 days

SydSteyerhart · x · 2026-11-22

Reacting to the claim that Terence Tao expected the Riemann Hypothesis to be proven around 2050 — and it actually took only 35 days — potterlab argues this fits the pattern of "jagged intelligence": next-token predictors intuitively shouldn't manage complex "special" thinking, but such thinking isn't categorically different from what we already expected AI to do.

Evidence keeps piling up that AI will keep getting more capable very quickly, and at each step we declare a new impossible milestone for it to smash through.

Original post →

More from AGI Musings

AGI Musings channel →