Superintelligent AI strategies won't beat smart diligent humans on values-heavy decisions
binarybits · x · 2026-11-15
In a thread with Raemon777, binarybits argues that on values-heavy questions, strategies from a superintelligent AI usually won't be dramatically better than those from a smart, diligent human.
His follow-up point: a usable strategy must be one the human leader understands well enough to defend to others. An AI advisor can nudge decisions at the margin, but only at the margin—any push toward something nonsensical will simply be rejected.
More from AGI Musings
- Superintelligent AI won't beat diligent humans on values-heavy decisions, argues binarybits — binarybits · 2026-11-15
- Commenter argues AGI automation is domain-specific, not a hockey-stick takeoff — jessi_cata · 2026-11-15
- tszzl credits open source: mechinterp research like Nick Cammarata's wouldn't exist without Qwen — tszzl · 2026-11-15
- zetalyrae's 'Taking AI Seriously': programmers' AI advice keeps retreating like a losing war — zetalyrae · 2026-11-15
- AGI feeling 'just a magic trick' from extra datasets and outsourced consultants, critic argues — sebpaquet · 2026-11-15
- Why big US AI labs are 'immortal': compute capex dwarfs opex, says analyst — teortaxesTex · 2026-11-15