AI Math's New Bottleneck: FAR Ranks Which Conjectures Deserve Compute
wellecks · x · 2026-09-11
As AI gets better at mathematics, deciding which problems deserve compute and mathematicians' attention becomes the harder problem—AI can attempt far more problems than experts can review. The FAR framework builds the missing ranking layer between the two stages:
- Starts from a research direction instead of a hand-picked conjecture, searching the literature for open problems;
- Attempts them at scale, then filters outputs so expert mathematicians only see the most promising cases;
- Validated in a combinatorics pilot.
The authors argue the "what should AI try to solve?" question will grow increasingly important as models improve.
Related event: CMU's FAR Framework Sifts 5,245 Papers for Conjectures Worth Solving(3 posts)→
More from AGI Musings
- Data engineering, not agent frameworks, is the real bottleneck for enterprise AI agents — dhruv2038 · 2026-09-11
- François Fleuret: Only Two Long-Term Futures — No Super AI, or Staying Fully Human With It — francoisfleuret · 2026-09-11
- IG reel debunking the 'winning the AI race against China' fallacy hits 500k likes — louisvarge · 2026-09-11
- Post-AI World Leaves No Room for Learning on the Job — rachittshah · 2026-09-11
- Researcher questions AI safety eval firm, citing 'blatantly sloppy' security and monitoring — Kyrannio · 2026-09-11
- AI researcher memes agent-swarm tinkering with He Jiankui's embryo-editing quote — dejavucoder · 2026-09-11