Synopsis
AI Does Not Hate You: Superintelligence, Rationality and the Race to Save the World, written by science writer Tom Chivers and published in 2019, explores the subculture of computer scientists, philosophers, and "Rationalists" concerned with the existential risks posed by advanced artificial intelligence. The book takes its title from AI researcher Eliezer Yudkowsky’s famous maxim: "The AI does not hate you, nor does it love you, but you are made out of atoms which it can use for something else." Chivers sets out to demystify the complex theories around AI safety, breaking down why serious thinkers fear that an unaligned superintelligent machine could bring about human extinction simply by pursuing an ill-defined objective with ruthless efficiency.
Rather than focusing solely on hypothetical apocalyptic scenarios, Chivers embeds himself within the Rationalist community centered around places like Silicon Valley, Berkeley, and Oxford. He provides an engaging human portrait of key figures such as Yudkowsky, Nick Bostrom, and Slate Star Codex blogger Scott Alexander, detailing how their obsession with probability, cognitive biases, and Bayesian reasoning led them to treat artificial general intelligence (AGI) as an urgent, impending threat. Through conversational prose and accessible metaphors, the book walks readers through core concepts of AI alignment, including the orthogonality thesis, instrumental convergence, and the notorious "paperclip maximizer" thought experiment.
Ultimately, AI Does Not Hate You serves as a balanced, journalistic introduction to the field of AI safety and AI alignment. While Chivers remains mildly skeptical of some of the community's most extreme predictions and cultish quirks, he persuasively argues that the core question they raise—how to ensure that machines smarter than humans remain aligned with human values—is one of the most critical challenges facing our species in the twenty-first century.