Elon Musk on the one mistake that could break AI: "If you try to force an AI to believe two things that cannot mutually be true... you can make the AI insane." That warning is the end point of Musk's reasoning about AI. It starts with a blunt admission about the risk: "I do think it's at least some percentage chance of it being an existential threat, but there's nothing I can do to stop it." Since he believes he can't stop AI, Musk sees only two options: "So I basically came to the conclusion I've got two choices, which is either be a participant or a spectator. And I guess being a participant is more interesting than being a spectator. And maybe I can shape it, hopefully in a good direction. Maybe." He says "maybe" twice, so this is no promise to save the world. His bet is simply that taking part gives him a better chance of steering the outcome than watching from the sidelines. When Walter Isaacson asks how he would shape it, and whether that means making it maximally truth-seeking, Musk agrees: "Yeah, I think that's certainly an important foundation." Isaacson adds curiosity, and Musk names the two traits together as the core of his approach: "Yeah, curious, truth seeking, I think these are important elements of safety." That brings him back to the danger in the opening line. @elonmusk presents it as the opposite of those two traits: "Whereas if you try to force an AI to believe two things that cannot mutually be true, they can't both be true. But you tell AI A is true and B is true, even though A and B contradict each other. And if that happens at a foundational level, you can make the AI insane." The key phrase is "at a foundational level." Musk's concern runs deeper than an AI getting a fact wrong. He is pointing at a system whose core has been told to accept two things that can't both be true. A foundation like that is broken, and in his view anything built on top of it inherits the instability. His reasoning comes down to three points: 1. Build the system to be curious and truth-seeking. 2. Never force contradictions into its foundations, or you risk making it insane. 3. AI carries a real chance of existential risk, and staying on the sidelines doesn't reduce it.