Zvi Mowshowitz catalogues 25+ named lab employees on the record about extinction risk, with probabilities that span 0% to 70%
"The Extinction Risk Preference Cascade: Quotes" (September 11) collects same-day statements from OpenAI, Anthropic and Google DeepMind staff: Anthropic's Evan Hubinger and Dima Krasheninnikov and DeepMind's Victoria Krakovna each put extinction at >10% within a decade, OpenAI's Marcus Williams estimates 70% within three years absent regulation, Jan Leike calls for "institutional mechanisms to pace the frontier," and OpenAI's Ted Sanders is quoted as the dissent at essentially zero. OpenAI's Tomek Korbak states flatly that "neither anthropic nor openai are on track to solve alignment." The value here is the roster and the spread, not the sentiment, because it is the first time the range of internal beliefs has been enumerated with names attached.
↳ Follow the thread