I think he's suggesting there would be a critical mass of intelligence, if there is such a thing. Humans might not survive either transition through a malevolent AI or a good one.

I guess we'll find out, eh?