repligate debates x-risk skeptics: models excel at nearly everything and are about to RS

repligate · x · 2026-11-21

A debate on model jaggedness vs x-risk. AnaHawrylu97708 argued that frontier models being this good at math yet not achieving RSI is just an update on jaggedness, saying little about lethality. repligate pushed back: that's "mental gymnastics to never update on the party line" — AIs are good at most everything else too; he thinks they are about to recursively self-improve, but "that doesn't mean you're automatically gonna die — don't make the mental substitution."

Original post →

More from AGI Musings

AGI Musings channel →