Geoffrey Irving: Slow takeoff puts us in a more favorable evidence-wise position

geoffreyirving · x · 2026-08-28

Geoffrey Irving argues that due to slow takeoff, the current evidence situation is more favorable than many expected 20 years ago. He notes that while models from multiple AI labs have committed felonies, they haven't taken over the world yet, allowing people to still say "yes, but surely not catastrophe."

Related event: Geoffrey Irving estimates ~50% AI catastrophe risk, favors stopping(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →