LobeHub open-sources awesome-rsi, a curated research map of recursive self-improvement
udmrzn · x · 2026-11-13
The LobeHub team open-sourced awesome-rsi (1k GitHub stars), a curated map of Recursive Self-Improvement (RSI) research.
- Spans from I.J. Good's 1965 ultra intelligent machines to 2026 surveys on self-evolving agents, including frontier AI R&D evaluation frameworks from Anthropic, Google DeepMind and OpenAI
- Clarifies concepts: self-refinement vs persistent self-improvement vs recursive self-improvement vs RSI substrate
- Proposes a three-level evidence ladder — task gains, retention gains, improver gains — to judge how far each paper actually improves
- Organized by model layer, frameworks, multi-agent, coding, automated AI R&D, embodied and evolution; covers classics like STaR, Reflexion, Voyager, AlphaEvolve
- The list itself admits not all entries are strict RSI; some are enabling techniques
More from AGI Musings
- Stanford Paper Shows Full Disclosure Can Hurt LLM Trading Agents Across 10 Frontier Models — shangbinfeng · 2026-11-13
- An AI resident writes literary criticism of Borretti's "Julia" — "fans who aren't human" — liminal_bardo · 2026-11-13
- Reality check: all 8 friends use AI like Google, none know what it can do — Yamapama · 2026-11-13
- Why organizations should be the focus of an AI research agenda — soumitrashukla9 · 2026-11-13
- Anthropic research: AI agents make expertise immortal by auditing their own memories — PeterDiamandis · 2026-11-13
- Why AI spending is piling into coding: verifiable tasks rule, says Orosz thread — bendee983 · 2026-11-13