OpenAI's Boaz Barak: human mathematicians stay essential even when AI proves theorems
burny_tech · x · 2026-09-24
OpenAI researcher Boaz Barak on what human mathematicians do in a world of mathematical superintelligence:
- "It will be kind of pointless if AI proves theorems that only AI can understand. So we want AIs to explain to humans."
- AIs will find it easier to explain their work to expert mathematicians than to the general public; those mathematicians will be the first to interact with AI, explore the right questions, and then jointly explain the results.
- Even if GPT-7 could explain just as well, he'd still rather hear Terry Tao's own perspective, and would still prefer reading books written by humans no matter how good AI writing gets.
Related event: Barak: AI Math Boom Exposes More People to Frontier Research(2 posts)→
More from AGI Musings
- Dario touts Claude-led gene editing discovery, but biologists call it overhyped — ccerrato147 · 2026-09-24
- AIDE² paper: self-improving research agent beats the version humans tuned for two years — egrefen · 2026-09-24
- Software engineers admit it: the cushiest job of all time may be over — anselm · 2026-09-24
- First-gen AI agents are like self-driving cars with steering wheels; an agent-native web is coming — SuB8u · 2026-09-24
- Logical Intelligence Claims First New 'Good Group' in 31 Years, AI-Assisted Topology Breakthrough — burny_tech · 2026-09-24
- Managing people feels like managing a team of agents, minus the emotions and quitting — zebird0 · 2026-09-24