Are the Foundations Strong Enough to Build More General Purpose Intelligence?

Are these foundations where you want to predict the next word and do auto complete strong enough to actually build, you know, systems that understand and form the basis for sort of more general purpose intelligence? Yeah, this is the big question. I think we don't know the answer with much confidence, but I think this is really driving what a lot of us in the field are trying to figure out. The field is arguably splitting a bit based on people's views on this question. It just seems like qualitatively these things are getting better quite quickly; there's no kind of obvious reason to believe that it would slow down.