Debating the '10% extinction chance' framing: attention now, credibility loss later

paulnovosad · x · 2026-09-11

Itaishér argues that the ">10%" p(doom) framing grabs attention short-term but erodes long-term credibility, and claims with unfounded specificity should be avoided in favor of other ways to communicate risk.

wwwojtekk quips that 10% is a good number—large enough to say "I told you so" after we all die, but below the 20% threshold where people start mocking failed predictions. The exchange captures a real tension in AI safety communication: alarmist precision vs. sober messaging.

Related event: Debate Erupts Over How to Frame AI Extinction Risk Probabilities(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →