Gregory's Server
These glorified autocomplete systems have zero chance of becoming intelligent because that's not what they're programmed for.
Nobody knows how to make something that's intelligent.
The "open letter warning about AIs being dangerous" was a deliberate distraction and PR stunt:
"Don't look at how we're screwing up all knowledge, look over here at this pretend threat. See how we're good people to be worried about something that we're pretending could harm you."