zetalyrae: AI governance is easier than alignment — rationalists would've scoffed a year ago

zetalyrae · x · 2026-09-04

zetalyrae notes that a year ago the median rationalist would have dismissed AI governance as impossible, preferring "lunatic technical alignment projects that will never work" — but governance is turning out to be easier.

Gregory Conti adds that policymakers have successfully acted against narrow economic interests before, and can do the same on AI. The exchange reflects a shift in AI safety circles from alignment-first toward governance routes.

Original post →

More from AGI Musings

AGI Musings channel →