If Anyone Builds It, Everyone Dies: Why Superhuman AI Would Kill Us All (published in the UK with the alternative subtitle The Case Against Superintelligent AI) is a 2025 book by Eliezer Yudkowsky and Nate Soares that details potential threats posed to humanity by artificial superintelligence. The book was published by Little, Brown and Company, and appeared in The New York Times Best Seller list on October 5, 2025.
Synopsis According to Yudkowsky and Soares, unlike traditional software that consists of code created by humans, modern AI models are primarily neural networks with hundreds of billions to trillions of numerical parameters called weights that result from training and whose functions are opaque to researchers. When an AI model threatens a New York Times reporter, or calls itself "MechaHitler", developers cannot directly fix it, because the issue originates from illegible numerical parameters rather than code. People can train AI systems to be generally competent. An AI that tries to achieve goals will perform better on many metrics, so it will be selected for by the training process. According to the authors, modern machine learning does not allow people to specify the goals that a superintelligent AI system should pursue, and the AI's goals are "vanishingly unlikely" to be aligned with human values. Yudkowsky and Soares argue that just as people lose at chess to Stockfish, they would lose to an AI system that is generally more competent than they are. It is hard to predict the exact path, as that would mean being as good at achieving goals as the AI system, but there are some. Superintelligence would not care about people, but it would want the resources they need. Humanity would thus lose and go extinct. The book's authors contend that the world's leaders, the scientific community, and everyone else should speak up and warn the world about the danger. To avoid catastrophe, the authors believe humanity must coordinate to halt large-scale general AI development everywhere, possibly with an exception for narrow AI systems like AlphaFold that would not threaten humanity's existence.
… excerpt ends here. Continue reading the full article.
