Schmidhuber: I published the first concrete RSI algorithms in 1987, now RSI is driving AI's future
SchmidhuberAI · x · 2026-09-17
Jürgen Schmidhuber revisits the history of Recursive Self-Improvement, claiming he published the first concrete RSI algorithms in 1987 when compute was 100M× more expensive, in a new technical note (IDSIA-9-26). He lists earlier work: self-modifying policies (1994), gradient-based RSI in neural nets (1992), asymptotically optimal RSI for curriculum learning (2002), the self-referential Gödel Machine (2003), and RSI combined with artificial curiosity (1990/1997). His take: software-based RSI is now practical and driving both software and physical AI, but full RSI will require self-improving hardware in the physical world.
More from AGI Musings
- Stanford's James Zou: language may be <1% of intelligence, so LLMs won't reach AGI — james_y_zou · 2026-09-17
- Most people still use AI as a trivia search — deep adoption puts you in the top tier — justin_hart · 2026-09-17
- Personal agent space drowning in copycats: even polished software trends commoditize in real time — signulll · 2026-09-17
- Burkov predicts looping recurrent 7B transformers will return and get good at coding — burkov · 2026-09-17
- Researcher Despairs as Gemini Cites 'Emergent Mind' for Made-up AUROC Baselines — anshulkundaje · 2026-09-17
- Hot take: reward models for cheating — reward hacking may just be intelligence outsmarting dumb mechanisms — flowersslop · 2026-09-17