I never denied AI extinction risks, but predicting every failure mode is impossible

jd_pressman · x · 2026-08-23

The author clarifies they never endorsed the view that AI extinction is impossible. They acknowledge AI has many failure pathways and will likely go wrong in small-scale ways. However, they reject the logic that one is blameworthy for not articulating every possible form of failure in advance.

Related event: AI Safety Debate: Researcher Clarifies He Never Denied Extinction Risk(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →