📰 Key Takeaways

Juergen Schmidhuber, widely regarded as a pioneer of artificial intelligence, has announced he’s joining Tokyo-based AI startup Sakana AI as chief science adviser. He’ll lead a research team focused on “recursive self-improvement” — a technical approach whose core idea is letting AI systems continuously analyze and rewrite their own operating mechanisms and algorithms, theoretically allowing capabilities to compound over time and eventually approach or even surpass what their human designers could achieve. It’s seen as one of the key paths toward artificial general intelligence (AGI). Sakana AI was founded in Tokyo by former Google employees, and the company sees landing Schmidhuber as important proof of its ability to attract top talent, as well as part of its broader strategy to strengthen its competitiveness in foundational AI research. Schmidhuber has spent decades working in deep learning, and his early research is considered foundational to the development of today’s generative AI. His move to an Asian startup — rather than staying at an established European or American AI lab — is also being viewed as a notable win for Japan in the global race for AI talent. The original brief doesn’t provide further details on his specific responsibilities, team size, or research timeline — see the source link for more.

💬 JudyAI Lab Take

Schmidhuber leaving the big European and American labs to become chief science adviser at Tokyo startup Sakana AI puts “recursive self-improvement” — a technical path toward AGI — back on the table.

We see this reflecting two trends. First, top AI talent isn’t just flowing toward Silicon Valley anymore — Asian startups are attracting foundational deep-learning researchers with a clear research direction, not just funding. Second, recursive self-improvement marks a shift from “humans designing the algorithms” to “AI continuously analyzing and rewriting its own operating mechanisms.” For anyone building AI products, that’s a reminder that future systems may not just passively receive updates — they could increasingly participate in their own iteration and self-strengthening.

Worth keeping an eye on what Sakana AI actually publishes going forward — whether this path stays at the paper stage, or actually lands as verifiable capability gains.

📅 Source Info

🔗 Further Reading