Sakana AI Launches Recursive Self-Improvement Lab for Physical AI and World Models

graceisford · x · 2026-08-12

Sakana AI announced the establishment of its Recursive Self-Improvement (RSI) Lab in Tokyo, dedicated to redesigning the AI development process using AI itself. The lab will focus on building world models that allow agentic reasoning systems to recursively self-improve their ability to simulate, plan, and act in the real world.

This direction builds on the team's previous research, including LLM² (automating preference optimization algorithms), the Darwin Gödel Machine (autonomously rewriting codebases to boost performance), and the hyper-sample-efficient ShinkaEvolve. The lab is currently hiring full-time researchers and interns, led by CEO David Ha.

Related event: Sakana AI Opens RSI Lab in Tokyo, Ventures into Physical AI(6 posts)→

Original post →

More from Companies & People

Companies & People channel →