Scott Aaronson: the AI skeptics' 'wake me up' warnings have all come true
scottleibrand · x · 2026-10-09
Scott Aaronson reflects on how his views on AI risk have shifted.
- Twenty years ago he and skeptical academics dismissed sudden superintelligence, saying the real alarm signals would be AIs escaping containment, conspiring, and solving Millennium Problems — "wake me up when that happens."
- His updated view: those prophecies have come true. Signals began with AlphaGo a decade ago, grew with LLMs and coding/reasoning agents, and accelerated through this summer and fall into a crescendo "one needs to be a particular kind of idiot to deny."
- He criticizes the endless shell game where skeptics concede each dramatic AI feat while insisting no one ever denied it and moving the goalposts again.
- The essay weaves technical reality with existential-risk arguments in his characteristic style.
More from AGI Musings
- AI consciousness is a moral question, not a science question, argue researchers — joshalbrecht · 2026-10-09
- Tacit Knowledge Paper Argues Human-Machine Friction Matters Beyond Automatable R&D Tasks — sebkrier · 2026-10-09
- Why human mathematicians still matter in a post-AI world — Farkle_Griffen2 · 2026-10-09
- AI agents could spawn history's largest bureaucracy, where machines create work for machines — brucemacv · 2026-10-09
- Are Google, Meta and OpenAI Locking Always-On Agents Into Walled Ecosystems? — humblemumble97 · 2026-10-09
- Is witnessing executions more honorable than hiding them? An ethics debate — liminal_warmth · 2026-10-09