Why smart people struggle to rebut AI doom arguments: multi-agent risks
Borg70955376 · x · 2026-09-18
A user argues that many ostensibly intelligent people have found AI existential risk analysis (e.g., Yudkowsky's position) worryingly hard to rebut well. Even without assuming a hard intelligence explosion, the strategic interactions among millions of agents more capable than humans are themselves a cause for concern. The "10% probability" mentioned in the discussion should not be read as a modal interpretation about possible worlds, but rather as a betting question: if you were betting against Yudkowsky on whether building AI kills everyone, what odds would you accept?
More from AGI Musings
- Claude Now Leads Over a Quarter of Anthropic's Own AI R&D Work — TansuYegen · 2026-09-18
- Andrew Ng calls AI doomer warnings from top labs "science fiction" — JFPuget · 2026-09-18
- AI Scores Just 21.2% on UN Development Statistics, Prompting Data Rebuild — TansuYegen · 2026-09-18
- Extreme Lab Lock-In: Frontier Labs Hoarding Best Models May Be the Underpriced Endgame — abhiadesai · 2026-09-18
- David Patterson: permanent mass unemployment will start by 2028, complete by 2030 — davidpattersonx · 2026-09-18
- Some Actors Will Always Escape Controls: Pre-AI Estonia Cyberattack as a Case Study — _onionesque · 2026-09-18