A new study aims to resolve lingering confusion over the meaning and interpretation of the "P(doom)" metric, a statistical measure used to quantify the probability of catastrophic global risks like artificial intelligence misalignment or nuclear war. Researchers argue that existing debates stem from differing assumptions about whether P(doom) represents a conditional probability—given certain conditions—or an unconditional probability reflecting baseline risk. The paper proposes a framework to standardize how experts and policymakers discuss existential threats, emphasizing the need for clearer definitions to avoid miscommunication in high-stakes discussions. Critics and proponents alike will likely weigh in on whether this approach could sharpen—or further complicate—efforts to assess and mitigate long-term risks.
