Sakana AI symposium: AI Scientist builds on Schmidhuber's 1987 self-improving AI concepts
SchmidhuberAI · x · 2026-10-29
An attendee recaps Sakana AI's symposium "From World Models to Real-World AI," featuring talks by Jürgen Schmidhuber and Sakana CEO David Ha.
The standout takeaway: Sakana's own AI Scientist (automated research system) is built on core concepts Schmidhuber formulated in 1987 — self-improving AI that learns to learn — while the world models lineage traces back to his 1990 work. Slides also showed his NNAISENSE 2014 motto "AI∀: AI For All," arguing AI gets 10x cheaper every 5 years and shouldn't be controlled by a few big companies.
An industry event linking the intellectual origins of meta-learning/self-improving AI directly to today's AI Scientist practice.
More from Companies & People
- Investor says Apple or Amazon should be allowed to acquire Instinct despite antitrust — johncoogan · 2026-10-29
- 10-person Spawn spends 30x payroll on AI tokens, ships 77,000 changes a month — mattshumer_ · 2026-10-29
- Dartmouth provost Santiago Schnell steps down over AI use in published work — docmilanfar · 2026-10-29
- GitHub Octoverse: 17.9B pushes this year, more than 2024 and 2025 combined — 0xkarasy · 2026-10-29
- 40 On-Device Gemma Projects Built in One Day at $60K Hardware Hackathon — agihouse_org · 2026-10-29
- Mistral throws launch parties in SF, NYC and London for ML4 while hiring across 7 cities — MistralAI · 2026-10-29