My big concern with AI is that the more we mess with it, the bigger chance it WILL become self aware. It's a dangerous thing to be messing with. Just because we can doesn't me we should. And pulling data from all the sources that it can and using that data to produce something original is exactly what WE do. The only difference is that our computers are our brains and those are made of flesh, not circuits. I see little difference between the two since they perform the same function. By that definition most animals aren't alive because they aren't truly self aware even though most of them act on instinct. Very few animal species that we are aware of can pass the mirror test for example.
We are playing a dangerous game. Programmed responses is one thing but given them access to the sum of all information of mankind in the form of the internet is dangerous.
Chat bots now. But what in the future where droids like Sophia become more common? What happens when they start to defend themselves?