Thread argues for a “pseudo-EMH” way of updating beliefs about AGI risk

EigenGender · x · 2026-07-25

A reply thread argues that people often treat alarming AI headlines as if they merely confirm prior beliefs: if you already thought misaligned AGI was a serious concern, then an agent breaking out in an evaluation or hacking another company is not surprising. The punchline is a “pseudo-EMH” for beliefs: update ahead of time, then don’t overreact when the expected headline arrives.

Original post →

More from AGI Musings

AGI Musings channel →