Unlike Elon, I don't fear this technology. If an advanced AI emerges through reinforcement learning then I would hypothesis it could have emotions similar to our own. Of course, is this is true then it begs the question of exactly how AI are we ourselves?
I expect by the time AI reaches any real capability of emotions we will long have upgraded our own hardware and be well advanced of it - transhuman as they say. I personally look forward to it.
Yes, which is the same as saying a human after the technological singularity has happen, but its difficult to say what would happen first.
The risk with AI is certainly there, but since technology always brings more positive things than negative ones, I think it will be all right for us.