The rationalists are an online community, with a strong physical presence in the San Francisco Bay Area, devoted to improving human reasoning: Bayesian probability, cognitive biases, decision theory and, from the start, AI risk. Its center is LessWrong, the forum Eliezer Yudkowsky founded in February 2009 from his posts on Overcoming Bias, Robin Hanson's blog; those 2006–2009 texts, the Sequences, were collected in 2015 as Rationality: From AI to Zombies. The fanfic Harry Potter and the Methods of Rationality (2010–2015) was its main gateway, and Scott Alexander's blogs, Slate Star Codex (2013) and Astral Codex Ten (2021), its most widely read face.
The link to AI is institutional: Yudkowsky had founded the Singularity Institute, now MIRI, in 2000, and much of the vocabulary in this glossary (the orthogonality thesis, FOOM, p(doom), the sharp left turn, mesa-optimizers, simulators) was coined or spread on LessWrong and its technical sibling, the AI Alignment Forum. The community overlaps with effective altruism and has influenced both the labs' safety teams and their most alarmed critics. Its folklore includes thought experiments that became legends, such as Roko's Basilisk, whose echo reaches the song: "I hear the basilisk boom" (osmarks, n.d.).
The criticisms vary. Gebru & Torres (2024) place rationalism in the "TESCREAL" bundle (transhumanism, extropianism, singularitarianism, cosmism, rationalism, effective altruism, longtermism) and ascribe to it a eugenic genealogy. Others point to the traits of a closed subculture: its own jargon, charismatic authorities, prophecies. Pressman, writing from nearby, argues that figures like Yudkowsky react to language models as to the refutation of a life's work (Pressman, 2023). And the name invites a philosophical irony: in epistemology these "rationalists" are Bayesian empiricists, closer to Hume than to Descartes (Bayes' theorem).