Schmidhuber says his RSI work dates to 1987, when compute was 100M times costlier
SchmidhuberAI · x · 2026-09-21
Jürgen Schmidhuber, replying in a thread with Hinton and LeCun, resurfaced his technical note tracing the lineage of Recursive Self-Improvement (RSI) research.
He claims to have published the first concrete RSI algorithms in 1987, when compute was 100,000,000x more expensive, and lists subsequent milestones: self-modifying policies since 1994, gradient-descent-based RSI in neural networks since 1992, asymptotically optimal RSI for curriculum learning since 2002, the mathematically optimal self-referential Gödel Machine since 2003, plus RSI combined with artificial curiosity and intrinsic motivation since 1990/1997 and recent work since 2020.
His core argument: with compute now cheap, RSI is driving the future of both software and physical AI.
Related event: Schmidhuber Reasserts He Pioneered Recursive Self-Improvement in 1987(4 posts)→
More from Research
- Researchers ask: why is there no AI-first scientific journal with AI-run peer review? — anshulkundaje · 2026-09-21
- Waypoint Bio argues AI in biology has a data problem before a model problem — anshulkundaje · 2026-09-21
- New paper estimates LLMs raised US natural unemployment rate by 0.1-0.2 points — mattbeane · 2026-09-21
- JevBench open-sources a benchmark for typed decision models scoring intelligence, calibration and speed — sull · 2026-09-21
- gLM2 designs a 2,500-AA polyketide synthase making a nylon precursor, 10x titer boost — anshulkundaje · 2026-09-21
- Skip the LLM for routing? Jev classifier ties Llama 70B in harness tests — 2BucChuck · 2026-09-21