Glonce

Sakana AI hires Jürgen Schmidhuber to lead recursive self-improvement research

businessThe Glonce editorsThe Decoder🔥70Glonce rating70Impact50Novelty60Relevance100Surprise65Players85Credibility80
Sakana AI hires Jürgen Schmidhuber to lead recursive self-improvement research

Sakana AI, a Tokyo-based AI startup, has hired Jürgen Schmidhuber as Chief Scientific Advisor, calling him the "father of modern AI." Sakana credits Schmidhuber with early work on world models in 1990 and with 1991 deep learning techniques that it says underpin much of today's AI boom. His 1987 thesis on meta-learning and recursive self-improvement, the idea that an AI system keeps improving itself, opened nearly four decades of research into how machines can learn to learn. Sakana says these ideas directly shaped its own projects, including the Darwin Gödel Machine and The AI Scientist. Schmidhuber will now help steer Sakana's newly announced RSI (recursive self-improvement) lab, which the company says aims to build a self-reinforcing research loop that makes machines smarter. To staff it, Sakana says it wants to assemble a "critical mass of world-class experts" in Tokyo. In a statement, Schmidhuber said Japan is the birthplace of foundational neural network architectures and advanced robotics, and called it a privilege to join Sakana AI to bridge those two worlds, adding that the future of intelligence is not just language but physical AI powered by world models, and that Sakana is positioned to build autonomous systems able to understand, simulate and interact with the physical world. The source material does not give a start date, contract length or compensation for the advisory role, does not name any prior employer or affiliation Schmidhuber is leaving, and does not name any other researchers hired for the new RSI lab.

Key facts

  • Sakana AI, a Tokyo-based startup, hired Jürgen Schmidhuber as Chief Scientific Advisor
  • Sakana credits Schmidhuber with 1990 work on world models and 1991 deep learning techniques, and with a 1987 thesis on meta-learning and recursive self-improvement that began nearly four decades of learning-to-learn research
  • Sakana says his ideas shaped its Darwin Gödel Machine and The AI Scientist projects
  • Schmidhuber will help lead Sakana's new RSI lab, aimed at building a self-reinforcing research loop to make machines smarter
  • Sakana says it wants to gather a "critical mass of world-class experts" in Tokyo for the lab

Why it matters

Schmidhuber is one of the field's most cited early researchers, and Sakana is framing his hire as validation of its bet on recursive self-improvement, the idea that AI systems can be built to keep making themselves smarter. Attaching his name to a formal RSI lab signals Sakana wants to be seen as a serious contender in that research direction, not just a smaller AI lab following larger rivals.

Who it affects

It affects Sakana AI itself, which gains a high-profile scientific figurehead for its research agenda, and the broader AI research community in Tokyo, where Sakana says it wants to recruit a critical mass of world-class experts for the new lab. It also matters to anyone tracking the lineage of ideas, like world models and meta-learning, that feed into current AI systems.

How to use it

This is a personnel and research-direction announcement rather than a product or service; there is nothing here to install, subscribe to or apply. Readers interested in Sakana's technical direction can watch for output from the new RSI lab.

How solid is it

The account comes from Sakana AI's own announcement and statements, including a direct quote from Schmidhuber, so the framing and credit given to his past work reflect the company's own characterization. No start date, contract terms, compensation, or names of other lab hires are disclosed.

Risks and caveats

The claims about the impact of Schmidhuber's 1990 and 1991 work, and how directly it shaped Sakana's own projects, are Sakana's own characterization rather than independently verified. Details on the scope, timeline, staffing or funding of the new RSI lab beyond Schmidhuber's appointment are not given.

“Japan is the birthplace of foundational neural network architectures and advanced robotics. It is a privilege to join Sakana AI to help bridge these two worlds. The future of intelligence is not just language; it is physical AI powered by world models. Sakana AI is uniquely positioned to reclaim Japan's legacy of innovation by building autonomous systems that can truly understand, simulate, and interact with the physical universe.”

— Jürgen Schmidhuber, Chief Scientific Advisor at Sakana AI