What happened
Sakana AI said Jürgen Schmidhuber is joining as chief scientific advisor and will lead its new RSI Lab, which studies recursive self-improvement — AI that improves itself.
Why it matters
The lab aims to create a cycle where scientific discoveries lead to further discoveries, raising machine intelligence; Schmidhuber's involvement signals the field is taking self-improving AI seriously.
What to watch
The company plans to develop a 'world model' that simulates how actions affect the physical world, for physical AI in manufacturing and robotics; whether this materializes hinges on progress in those areas.
WHO IT HITSAI research teams working on self-improving systems and physical AI for manufacturing and robotics may see new competition or collaboration opportunities, as Sakana AI gains a high-profile advisor and sets its sights on world models.
Summaries like this, in your inbox every morning.
Sakana AI, a Japanese AI startup, has brought on Jürgen Schmidhuber, a researcher known for co-developing LSTM networks, as chief scientific advisor. He will guide a new lab dedicated to recursive self-improvement — a field exploring how AI can improve itself. This move connects Sakana AI to a long history of AI research, as Schmidhuber's 1987 thesis on self-improvement and meta-learning is cited as an influence on modern projects like the Darwin Gödel Machine.
The company's ambitions extend beyond software. It plans to develop world models that simulate the physical effects of actions, with the goal of building physical AI for manufacturing and robotics. Japan's strength in both neural network foundations and advanced robotics is something Schmidhuber highlighted, and he said he will visit Tokyo regularly to work with the team.
The stakes hinge on whether Sakana AI can turn these research directions into practical systems. If successful, it could carve out a niche in physical AI, but the timeline and commercial viability remain uncertain.
For example, today's edition would include:
AI-summarized, only the topics you pick: one digest a day via Email, LINE, or Slack.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
On 2026年9月15日, TypeSafe AI published System One Model and opened early access to Jev, which returns only type-…
Anthropic said Claude largely on its own found an unknown enzyme system, "ART," in DNA databases
Sakana AI hired Jürgen Schmidhuber as Chief Scientific Advisor, crediting his 1990 world-model work, 1991 deep…
OpenAI is preparing to preview GPT-6 Cyber in the coming weeks, with a limited number of Daybreak Red program…
The author gave Instinct, an invite-only AI agent reachable via iMessage and WhatsApp, access to her email and…
Google's Gemini 3.8 Live update adds a "Live Avatar" that lip-syncs and shows facial expressions in real time
