Microsoft researcher cracks 31-year-old KLS conjecture with AI agents exploring 100+ proof paths
rohanpaul_ai · x · 2026-10-22
A Microsoft researcher and collaborator resolved the 31-year-old Kannan-Lovász-Simonovits (KLS) conjecture in a 140-page proof, using human-directed AI search: agents explored over 100 research trajectories, ran derivations, prepared the manuscript and verified proofs, while the researchers pruned weak directions and contributed the recursive method and error bound. Intuitively, KLS says a random walker inside a convex body can't be trapped by hidden narrow passages and can explore it efficiently even in huge dimensions.
Related event: Microsoft researchers crack 31-year-old KLS conjecture with human-led AI(3 posts)→
More from AGI Musings
- Defenders' AI, not attackers, found the vulns — fueling AI offense-defense debate — cephaloform · 2026-10-22
- 'Agent + cybersecurity will be HUGE': devs debate returning to security careers — max_paperclips · 2026-10-22
- Demis Hassabis: today's AI will look like nothing compared to what's coming in 1–2 years — VraserX · 2026-10-22
- AI Security Enters a Machine-Speed Arms Race, Human Patch Cycles Are Dead — srimisra · 2026-10-22
- Box CEO: Cybersecurity Will Be Where AI Creates the Most New Jobs — rakyll · 2026-10-22
- Digital minds don't live and die with each instance: a case for pattern-based identity — RileyRalmuto · 2026-10-22