If Anyone Builds It, Everyone Dies, by Eliezer Yudkowsky & Nate Soares

If Anyone Builds It, Everyone Dies: Why Superhuman AI Would Kill Us All (2025)

The argument of AI doomers taken to its full extent. Emblematic of both the content and style of the existential-risk faction in Silicon Valley. I think they are wrong, and wrong in an instructive way: an argument pitched at extinction crowds out every institutional question that is actually live right now. But read it, because the register tells you as much as the reasoning does.

Read it asking: if the risk really is extinction, why is almost none of the proposed remedy democratic? Notice how the quickly the logic of existential risk arrives at rule by the people who understand it best.