As agents take on longer-running tasks, human value on high-leverage decisions goes up

morqon · x · 2026-11-19

Countering the argument that "a thing definitionally better than you at everything is better than you at everything," trq212 offers a counterintuitive observation: as agents handle longer and longer-running tasks, the value of working with them on high-leverage decisions goes up, not down — human judgment at critical junctures matters more as models get stronger and tasks stretch longer.

Original post →

More from AGI Musings

AGI Musings channel →