Sakana AI Expands RSI Lab for Embodied AI Self-Improvement

hardmaru · x · 2026-08-12

Sakana AI announced the expansion of its Recursive Self-Improvement (RSI) Lab to bridge RSI with physical AI and robotics. Led by CEO David Ha, the lab aims to transition from human-led R&D to autonomous, self-improving intelligence engines.

The research spans frontier language models, world models that learn physical dynamics from real-world interaction, and the recursive self-improvement loop binding them. This builds upon early "world models" research by Ha and Jürgen Schmidhuber, where agents learned policies inside generated "dreams."

Related event: Sakana AI Establishes RSI Lab for Recursive Self-Improvement and Robotics(4 posts)→

Original post →

More from Companies & People

Companies & People channel →