Scott Aaronson launches new UT Austin course on AI Alignment Theory
burny_tech · x · 2026-10-05
Quantum computing scholar Scott Aaronson is teaching CS395T "AI Alignment Theory" at UT Austin this semester. The course focuses on theoretical and mathematical foundations of aligning and controlling powerful AI systems—design, training, monitoring, and scaffolding—explicitly noting the field lacks a canonical body of results. Students will read and debate the most influential conceptual and mathematical works in alignment, before and during the LLM era, with presentations and projects central. Aaronson frames it as his long-delayed engagement with Yudkowsky-style ideas from 20 years ago.
Related event: Scott Aaronson Launches AI Alignment Theory Course at UT Austin(2 posts)→
More from Companies & People
- Consultants' 'independent' cloud advice is tilted by reseller margins, referral fees, and partner-tier incentives — DavidLinthicum · 2026-10-05
- Fintech AI deployment nearly paid out a Tesla claim for a Camry — Ornery-Length- · 2026-10-05
- Reddit post claims OpenAI fired a safety researcher three weeks after a tweet — Puzzleheaded-King584 · 2026-10-05
- Bezos: Never hire your friends before 40, only hire them after — rohanpaul_ai · 2026-10-05
- Physical AI field session in Bengaluru to tackle post-deployment evals and retraining loops — carrycooldude · 2026-10-05
- threepointone urges AI company devrels to drop performative posting and be themselves — threepointone · 2026-10-05