Sudden catastrophe versus gradual harm in AI risk

A post argues that Dan Selsam's letter fails to grapple with AI endangering 10,000 people. The writer sees its apparent focus as AI remaining unmonitored until it is sure it can overpower humanity in one shot, and expresses uncertainty about how to think about that framing. Another post shares the concern about risks being conceived as a “binary instant apocalypse.” It suggests gradual danger going unnoticed until too late may be more plausible, but also questions whether either scenario seems plausible.