Hinton's Gödel lecture: AI is a tiger cub we must ensure never wants to kill us
mmbronstein · x · 2026-10-24
Geoffrey Hinton gave the Gödel Lecture at TU Wien, where a slide compared humanity's AI situation to raising a tiger cub: cute now, but fully capable of killing you when grown. Survival offers only two options — get rid of the cub (not an option for AI), or find a way to ensure it never wants to kill you. It continues his alignment warning: the challenge is not capping capability but removing the will to harm.
More from AGI Musings
- 42 decacorns created in 12 months vs ~20 per decade for 50 years — alvelda · 2026-10-24
- Sam Altman suggested AI labs merge if they lose control — critics push back — robleclerc · 2026-10-24
- Refine founder pushes back on AI peer review hype: science is not naturally aligned — JessicaHullman · 2026-10-24
- Kids are building games with AI before they can read or write full sentences — thisiskp_ · 2026-10-24
- AI is far better at finding vulnerabilities than writing code, so software costs will rise — mgill25 · 2026-10-24
- Microsoft chief scientist Eric Horvitz: AI bubble may mirror 1847 railway boom, infrastructure wins — erichorvitz · 2026-10-24