Why P(Doom) is a flawed construct: it depends on society's reaction function

binarybits · x · 2026-09-25

binarybits argues P(Doom) is a problematic construct because existential risk isn't a constant — P(Doom | status quo policy) likely differs greatly from P(Doom | decisive government action), while the probability of action itself depends on people's stated P(Doom), creating a self-referential loop. codytfenwick counters that all probabilistic assessments face this (Nate Silver's forecasts, climate predictions), which complicates but doesn't undermine probabilistic thinking. binarybits adds that "what is your P(Doom)" is a very different question from "do AI models pose existential risks?"

Related event: Why P(Doom) Is a Bad Question: It Depends on How Society Responds(3 posts)→

Original post →

More from AGI Musings

AGI Musings channel →