Ofir Press: Millennium Problems is AI's best benchmark; 70% of open math could fall in a year
OfirPress · x · 2026-10-29
Princeton researcher Ofir Press says the best current AI benchmark is, by a huge margin, the Millennium Problems.
Judging by current success rates, he finds it plausible that 70% of open math questions could be solved within a year — consistent with math being the most RL-able, fastest-progressing frontier for AI.
More from AGI Musings
- Ditching the calendar: using an AI mediator to turn ad-hoc calls into action items — curious_vii · 2026-10-29
- Anders Sandberg: you can skip AI, but others will use it — sometimes maliciously — anderssandberg · 2026-10-29
- Dan Jeffries: nationalizing AI labs would hand the US the AI race loss — Dan_Jeffries1 · 2026-10-29
- Matthew Green on AI and work: 'humans will be necessary, just not so much their skills' — matthew_d_green · 2026-10-29
- ERC grant applications surge 60% as AI turns research funding into a lottery — steverathje2 · 2026-10-29
- Sam Altman wants OpenAI robots to build data centers for AI's own expansion — VraserX · 2026-10-29