Sakana AI announced the hiring of Jürgen Schmidhuber, the inventor of deep learning and world models, as its Chief Scientific Advisor. Schmidhuber is credited with foundational work in deep learning techniques from 1991, which underpin much of today's AI advancements.
Schmidhuber's 1987 thesis on meta-learning and recursive self-improvement laid the groundwork for nearly four decades of research into how machines can learn to learn. Sakana AI claims these ideas directly influenced its projects, including the Darwin Gödel Machine and The AI Scientist.
Japan, the birthplace of foundational neural network architectures and advanced robotics, is a privileged setting for Schmidhuber to help bridge these two worlds. Sakana AI aims to reclaim Japan’s legacy of innovation by building autonomous systems that can truly understand, simulate, and interact with the physical universe.
"The future of intelligence is not just language; it is physical AI powered by world models," said Jürgen Schmidhuber, Chief Scientific Advisor. Sakana AI plans to use his expertise to steer its new RSI lab, which aims to build a self-reinforcing research loop that makes machines smarter.
The announcement follows Sakana AI's recent focus on autonomous systems and world models. The company did not specify the timeline for new projects but emphasized its goal to gather a 'critical mass of world-class experts' in Tokyo.
Source: thedecoder
