Tokyo-based Sakana AI has hired Jürgen Schmidhuber as Chief Scientific Advisor, announcing the appointment on September 24, 2026. The startup describes him as a foundational figure in modern AI, highlighting his early research on world models in 1990 and deep learning techniques in 1991.
Schmidhuber’s 1987 thesis explored meta-learning and recursive self-improvement: how machines can learn to learn and improve their own capabilities. That work began a research program spanning nearly four decades. Sakana says his ideas directly influenced its Darwin Gödel Machine and The AI Scientist projects.
Explaining his decision to join, Schmidhuber emphasized Japan’s contributions to foundational neural network architectures and advanced robotics. He sees an opportunity to connect those fields at Sakana, with intelligence extending beyond language to physical AI driven by world models. His stated ambition is to develop autonomous systems capable of understanding, simulating and interacting with the physical world, while helping renew Japan’s role in technological innovation.
Schmidhuber will help guide Sakana’s new RSI lab, which seeks to establish a self-reinforcing research loop that improves machine intelligence. To support that effort, the company plans to bring together a substantial concentration of leading experts in Tokyo.