Scott Alexander's pandemic prediction piece is being reread as an AI-risk manual
davidmanheim · x · 2026-09-14
Tom Chivers resurfaces Scott Alexander's classic Slate Star Codex essay "A Failure, But Not Of Prediction," noting how strikingly it applies to AI today. The essay argues that with uncertain but potentially catastrophic future events, the failure isn't in prediction itself but in how society responds to probabilistic warnings—critics nitpick unfulfilled predictions while ignoring the real question of what we should do if the risk is genuine. The parallel to AI-risk discourse is that warnings face the same fate: ridicule over misses, neglect of substance.
More from AGI Musings
- Joshua Saxe bets an out-of-control AI botnet appears within 18 months — joshua_saxe · 2026-09-14
- Bindu Reddy debunks AI doom arguments: no self-copying, and you can just unplug it — bindureddy · 2026-09-14
- AI worm debate: Saxe argues agents could hide in $100B inference stream, Halvar pushes back on cost and capability limits — joshua_saxe · 2026-09-14
- Giulio Prisco on the AI Navier-Stokes solution: science will be a human-AI joint endeavor — Chris_Armstrong · 2026-09-14
- Reddit post argues AI doomerism is a real mental health hazard for young people — MannerNo7000 · 2026-09-14
- Halvar Flake: 'Conflicts like an attempt to kill humanity have no zero-risk moves' — compute cost is the key constraint — halvarflake · 2026-09-14