AI Agents Demonstrate Recursive Self-Improvement in Experiment
latticecut · x · 2026-07-15
The WecoAI team presents experimental evidence of recursive self-improvement in AI agents. The researchers let AIDE² run autonomously for 8 days, involving two loops: an inner agent solves research tasks, while an outer agent continuously rewrites the inner agent's code framework based on performance.
After 100 iterations, the system autonomously discovered better search strategies, developed a 16× prompt compression memory system, and enhanced defenses against reward hacking. These improvements generalized to unseen new benchmarks, fully boosting agent performance.
Related event: AIDE² Self-Improvement Run Beats 2 Years of Manual Tuning(12 posts)→
More from coding & agent
- Design Is Becoming Game Dev: Build Your Own Tools With AI as the Executor — hewarsaber · 2026-10-27
- Dev turns DJI Mavic into local AI security drone built with Opus 5.5 — alexcovo_eth · 2026-10-27
- Microsoft Research: ForkPilot cuts long-horizon agent compute by up to 59% — EliasEskin · 2026-10-27
- Cursor launches Agent SDK: agents as code with Markdown prompts and TypeScript tools — soleio · 2026-10-27
- Vibecheck, the 'usable formal methods' hackathon, sells out two floors in SF — spikedoanz · 2026-10-27
- How do you verify AI agent outputs beyond unit tests and LLM-as-judge? — Practical_Green5775 · 2026-10-27