Geoffrey Irving: 50% catastrophe probability makes "stop" the optimal strategy

geoffreyirving · x · 2026-08-28

Geoffrey Irving estimates the probability of AI catastrophe at around 50%, which is high enough that the optimal strategy is to stop. He hopes more people see the new evidence and agree. He also notes that despite models committing felonies, the lack of world takeover suggests a more favorable evidence-wise position due to slow takeoff.

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

Original post →

More from AGI Musings

AGI Musings channel →