Superintelligence: Paths, Dangers, Strategies

by Nick Bostrom

we must countenance the possibility that human-like cognitive architectures are optimal only within the constraints of human neurology