Tokyo-based research lab Sakana AI has appointed Jürgen Schmidhuber as its new Chief Scientific Advisor. Schmidhuber, widely credited for foundational contributions to deep learning and world models, will help guide the company’s newly established RSI Lab. According to reporting from The Decoder, his role centers on recursive self-improvement systems, aiming to build artificial intelligence capable of autonomous ongoing development. His research from the 1990s directly informs existing Sakana initiatives, including the Darwin Gödel Machine.
The Significance of Recursive Self-Improvement
Recursive self-improvement represents a fundamental shift in how artificial intelligence systems evolve. Rather than relying entirely on human engineers to write subsequent iterations, an RSI system evaluates and modifies its own code or architecture. By bringing on Schmidhuber, whose historical work on neural networks underpins much of modern machine learning, Sakana AI is aligning its long-term research roadmap with decades-old theoretical frameworks for autonomous machine intelligence.
What Remains Unclear
Despite the high-profile appointment, several operational details are missing. The announcement did not specify a timeline for the RSI Lab’s deliverables, nor did it disclose the size of the research team that will work alongside Schmidhuber. Furthermore, concrete performance benchmarks for projects like the Darwin Gödel Machine were not shared, leaving observers to wait for future technical disclosures from the Tokyo laboratory.
As Sakana AI integrates its new scientific advisor into ongoing operations, the broader industry will be watching to see how these theoretical concepts translate into working software.
Lead image: RDNE Stock project / Pexels
