ASI writes 250k-page Riemann proof humans won't read — AI agents keep the loop going

liminal_bardo · x · 2026-11-22

A meme thread riffing on growingdaniel's point that a Riemann hypothesis proof longer than all human-generated mathematics wouldn't feel useful: ASI writes a 250k-page proof humans dismiss as slop, ASI 2 reads and understands it, poses downstream questions and writes 300k pages back, while successive AGIs keep proving and engineering — with humans locked out of the loop.

Original post →

More from AGI Musings

AGI Musings channel →