emergence
I don't think that the scaling hypothesis gets recognized enough for how radical it is. For decades, AI sought some kind of master algorithm of intelligence. The scaling hypothesis says that there is none: intelligence is the ability to use more compute on more data.
