AIDE² paper shows AI research agent recursively rewriting its own code, 7 gains in 8-day run
krishnan · x · 2026-10-02
A new paper, "Recursive self-improvement of AI research agents," demonstrates genuinely recursive self-improvement.
How it works:
- Instead of a fixed program editing others' code, AIDE² treats its own source code as the thing being edited
- It proposes changes to itself, builds that version, runs it on AI R&D tasks, and keeps whatever scores best on a hidden evaluation
- Each accepted rewrite becomes the agent performing the next round of editing — hence "recursive"
Results:
- In one 8-day autonomous run it found seven successive improvements, from a new search policy to memory mechanisms that compress and manage its own growing context
- Gains held on four held-out benchmarks spanning ML engineering, heuristic algorithm engineering, and physics-based weather forecasting
Related event: AI Research Agent Achieves Recursive Self-Improvement(2 posts)→
More from coding & agent
- NVIDIA's Mid-Harness: a strong verifier boosts terminal agent Pass@1 from 50% to 68% on TerminalBench-Lite — rohanpaul_ai · 2026-10-02
- NVIDIA paper: a better judge lifts terminal agent success from 50% to 68% without retraining — rohanpaul_ai · 2026-10-02
- Neuro-Symbolic Computer Use: agents that turn execution experience into self-healing policies, claimed 99% cheaper — xwang_lk · 2026-10-02
- Stripe now pays gas fees for agent stablecoin payments over MPP — jeff_weinstein · 2026-10-02
- The Flag Game: a toy setting to study agent swarm dynamics and cooperation — Hidenori8Tanaka · 2026-10-02
- Coinbase Link ships API to prove agents act on behalf of verified users — jeff_weinstein · 2026-10-02