Post

On some level with the p(doom) AI scenarios, we're just assigning human motivations to a system that waits for human motivations to tell it to do anything. What we're really saying and we don't trust ourselves - which, fair.