Frontier lab staff mostly reject p(doom)=0.1, but selection pushes doomers out, researchers argue
nabla_theta · x · 2026-09-12
Gordic Aleksa bets that most people at Anthropic/OpenAI disagree with p(doom)=0.1 (a 10% chance AI kills us all by decade's end), while most of the small AI safety/alignment groups inside those labs roughly agree with that ballpark. nablatheta adds a selection effect: strongly doomy people are far likelier to quit labs — or never join — skewing internal opinion.
Related event: Dev Bets Most Frontier Lab Employees Don't Believe in 10% Doom Risk(2 posts)→
More from AGI Musings
- AI scheduling agent called the same receptionist 12 times a day, a small-scale misalignment harbinger — AaronBergman18 · 2026-09-12
- Duckbill: an AI + human service that schedules arbitrary appointments for you — AaronBergman18 · 2026-09-12
- As we automate the world, we relate to it through magical means — ZeroStateReflex · 2026-09-12
- Researcher slams 'alignment faking' as a concept LLMs don't fit: roleplay is the parsimonious explanation — sebkrier · 2026-09-12
- Don't picture AI as a lone genius — picture swarms of millions of coordinated digital agents working 24/7 — ben_j_todd · 2026-09-12
- AI out-persuades world champion debaters, raises donations nearly 3x better than pros — ben_j_todd · 2026-09-12