Misjudging AI capability — and our assumption we know better — makes alignment harder
snowclipsed · x · 2026-09-28
The author argues that a core human flaw is failing to recognize when we may be poor estimators of AI capability, while being extremely assumptive that we know better. This same bias, they contend, is also part of what makes alignment hard.
More from AGI Musings
- WSJ Goes Inside the Subculture Obsessed With AI Doom Long Before Everyone Else — sapinker · 2026-09-28
- New NBER Paper Finds No Evidence of AI-Driven Unemployment Among Recent College Grads — emollick · 2026-09-28
- 'Wanted to build god, ended up building a secretary that can code' — repligate · 2026-09-28
- If a frontier lab admits its AI can't be contained, that lab should be shut down — kevinnbass · 2026-09-28
- Does immediate Lean formalization make discovering new math harder? — LucaAmb · 2026-09-28
- 3Blue1Brown's HackMIT keynote asks: are you the collie or the pug? — AlexKontorovich · 2026-09-28