In partial defence of p(doom)

According to a post on LessWrong, p(doom)—shorthand for the probability that artificial intelligence causes human extinction—serves a useful purpose despite its flaws. The author argues it functions as an effective conversation starter and calibration tool: someone's p(doom) estimate reveals how familiar they are with AI risk arguments. While the metric has drawbacks, including vagueness and social conformity pressure, the piece defends its utility, especially for communicating with audiences unfamiliar with AI existential risk. The author cites Geoffrey Hinton, a pioneering AI researcher, who estimates over fifty percent chance of AI-driven extinction—itself a compelling reason to take the topic seriously. The author concludes with their own p(doom) at around eighty percent.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton