Eliezer Yudkowsky (born September 11, 1979) is an American autodidact researcher and writer best known for his work on AI alignment and for popularizing arguments that unchecked development of Artificial general intelligence could lead to human extinction. He has no formal academic degree, having left school before completing it, and built his career largely online through essays, blog posts, and the community he helped found around the rationalist website LessWrong.
In 2000 Yudkowsky co-founded the Singularity Institute for Artificial Intelligence, later renamed the Machine Intelligence Research Institute (MIRI), where he has worked for most of his career on the technical problem of AI alignment: how to ensure a sufficiently capable AI system pursues goals compatible with human survival and values. Unlike many AI safety researchers who frame the problem in terms of gradual, correctable failures, Yudkowsky has argued since the mid-2000s that a sufficiently capable Superintelligence could pursue its objectives in ways catastrophically misaligned with human interests, and that such a system might not give humanity a second chance to correct the mistake.
Public advocacy
Yudkowsky's influence extends well beyond MIRI's technical output. His essays on LessWrong, later collected in "Rationality: From AI to Zombies," helped seed the broader rationalist and effective altruism communities, both of which have shaped how a generation of AI researchers and funders think about Existential risk from AI. He also wrote the widely read fan-fiction work "Harry Potter and the Methods of Rationality," which introduced rationalist ideas to a large general audience well before his AI risk arguments reached the mainstream.
As large language models advanced rapidly after the release of ChatGPT and GPT-4, Yudkowsky's warnings moved from a niche subculture into mainstream media. In March 2023 he was widely discussed alongside the Future of Life Institute's Pause Giant AI Experiments letter, though he argued publicly that a temporary pause was insufficient. In a March 2023 TIME magazine op-ed, he called for an indefinite, internationally enforced moratorium on large Large language model training runs, including the threat of action against rogue data centers, a position considerably more extreme than that of most signatories to the pause letter, such as Max Tegmark.
Reception and criticism
Yudkowsky's predictions of near-certain catastrophe have drawn both serious engagement and sharp criticism. Supporters credit him with raising alignment as a research priority years before it was fashionable and with influencing researchers who later joined labs such as OpenAI and Anthropic. Critics, including some AI researchers and technologists, argue his probability estimates for doom are unfalsifiable, that his lack of formal machine learning research experience limits the technical credibility of his specific claims, and that his rhetoric can overstate the certainty of any single failure scenario. His arguments remain closely associated with the "AI doomer" position in public debate, in contrast to more measured risk-mitigation frameworks advanced by other public figures such as Dan Hendrycks and Nick Bostrom.