P(doom)
A subjective probability estimate used in AI safety research to quantify the likelihood that artificial general intelligence could lead to catastrophic or existential outcomes. Individual researchers assign specific numerical values based on personal judgment. The metric serves as an informal tool for discussing existential risk from advanced AI systems.
Timeline (last 90 days)
P(doom) refers to the probability that AI will destroy humanity, usually on a ten-year horizon; for the average American the median is close to zero, while in AI labs it is at least in the double digits.
P(doom) refers to an informal metric used by AI safety researchers to quantify existential risk, representing a subjective probability estimate that AI will cause a catastrophe, with no unified scientific calculation method.
AI researcher Eliezer Yudkowsky has publicly stated that his personal P(doom) estimate exceeds 90%.
Researcher Paul Christiano has given a relatively conservative estimate of P(doom).
All Facts (4)
P(doom) refers to the probability that AI will destroy humanity, usually on a ten-year horizon; for the average American the median is close to zero, while in AI labs it is at least in the double digits.
50%UnverifiedP(doom) refers to an informal metric used by AI safety researchers to quantify existential risk, representing a subjective probability estimate that AI will cause a catastrophe, with no unified scientific calculation method.
50%UnverifiedAI researcher Eliezer Yudkowsky has publicly stated that his personal P(doom) estimate exceeds 90%.
50%UnverifiedResearcher Paul Christiano has given a relatively conservative estimate of P(doom).
50%