Iliad Research Starts 'Textbook From the Future' With a Dozen Researchers to Fix AI Alignment's Missing Textbook
burny_tech · x · 2026-10-06
Responding to Scott Aaronson's observation that AI alignment still lacks a textbook, Iliad Research has launched "Textbook from the Future," with about a dozen researchers working on it.
Published project abstracts cover theoretical foundations for alignment: simplicity priors and learning (how far modern LLMs approximate Solomonoff induction, and how degeneracy underpins simplicity), universal predictors (MDL, logical inductors, resource-bounded reasoning), and causality — aiming to ground prediction, calibration, and reasoning in compression, algorithmic probability, and logical uncertainty. More to come, the team says.
More from Safety
- NY Assembly Member Accuses OpenAI of Perjury Over Flipped RAISE Act Testimony — DavidSKrueger · 2026-10-06
- Cisco Benchmarks Decision Models: Jev Nears 31B LLM Judge on Zero-Shot Safety Classification — aminkarbasi · 2026-10-06
- Hugging Face launches Open Alignment team for open-model safety — aidangch · 2026-10-06
- Report: China's ASML DUV access drives AI chipmaking; call to ban exports and servicing — teortaxesTex · 2026-10-06
- Meta crawler hammers one site with 700k requests/day; Cloudflare adds x402 pay-per-crawl — kleffew94 · 2026-10-06
- Review essay asks who really has power to shape global AI governance institutions — ProfChesterman · 2026-10-06