Superintelligence: Paths, Dangers, Strategies

by Nick Bostrom

A 10% probability of HLMI not having been developed by 2075 or even 2100 (after conditionalizing on “human scientific activity continuing without major negative disruption”) seems too low.