Superintelligence
If intelligence becomes scalable, control becomes the central political problem.
The 2014 book that moved existential risk from internet forums into boardrooms — once Musk, Gates, and Altman praised it publicly, the safety debate had a canonical text. Bostrom’s core case stands apart from the hype cycles that followed it: if capability outruns alignment, control becomes the hard problem, and it should shape architecture before scale.