Jürgen Schmidhuber will lead the development of agent-native world models and recursive self-improvement for physical AI at a newly established research center in Tokyo. Sakana AI appointed the computer scientist as its Chief Scientific Advisor to oversee the RSI Lab and trigger a compounding cycle of discovery to improve machine intelligence. The company is assembling a group of world-class experts in Japan to realize these goals.

Schmidhuber is known as the father of modern AI and pioneered work in the 1990s on meta-learning and deep learning that informed Sakana AI's prior projects, such as The AI Scientist and the Darwin Gödel Machine. He has researched how machines can learn to learn for 40 years, creating frameworks for world models before computing costs became practical for large-scale implementation.