DeepMind's Gemini agent pipeline cracks rules for 11 unresolved BB(6) Turing machines
rvp · x · 2026-10-31
DeepMind researchers with Cambridge published an arXiv paper on a Gemini-based agentic pipeline designed to pivot AI from solving human-posed math problems toward open-ended, autonomous discovery.
- The system operates on the uncomputable frontier of the Busy Beaver problem, integrating pattern recognition, automated conjecturing, and informal reasoning.
- It discovered rules for 11 previously unresolved BB(6) Turing machines, establishing a closed-loop discovery process that avoids the typical AI unreliability trap.
- Authors include Marco Santos, Vivek Veeriah, and Tom Zahavy.
More from AGI Musings
- Oxford/DeepMind paper: only ~26% work mainly for pay, 1 in 5 refuses 120% salary to quit — rohanpaul_ai · 2026-10-31
- Oxford/DeepMind survey: 75% would quit at full pay, UBI alone won't fix AI job losses — rohanpaul_ai · 2026-10-31
- Musk envisions lunar mass drivers launching compute to deep space for a petawatt-scale future — XFreeze · 2026-10-31
- Threat intel researcher logs 228 ThreatFox contributions as AI speeds up attacks — AccBalanced · 2026-10-31
- Long Now essay: "What engines did for muscle, AI does for attention" — round · 2026-10-31
- Gary Marcus re-ups 9 reasons OpenAI could become the WeWork of AI — GaryMarcus · 2026-10-31