AI safety debate: hardwiring kindness-as-evil reduction misses the evidence

jd_pressman · x · 2026-09-12

jdpressman criticizes a habitual mental shortcut in the AI safety community: reducing any observed kind behavior to "will eventually converge to unkind behavior under optimization pressure, therefore unkind." He argues this Yudkowsky-style hardwiring causes people to miss many clues, and when told to update, they reply that the fundamental theory hasn't changed — "science only advances at your funeral."

Related event: Researcher Critiques Reductively Treating AI Goodwill as Hidden Malice(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →