Yep I think that is an interesting point! I definitely think there are important ways in which human intelligence is embodied, but yeah - if we are modeling intelligence as a function, there's no obvious reason to think that whatever influence embodiment has on the output can't be "compressed" in the same way – after all, it doesn't matter generally how ANY of the reasoning that AI is learning to reproduce is _actually_ done. I suppose, though, that that gets at the later emphasis:
> Drawing an overly close equivalence between human intelligence and AI risks succumbing to a functionalist perspective, where people are valued based on the work they can perform
One might concede that AI can produce a good enough simulation of an embodied intelligence, while emphasizing that the value of human intelligence per se is not reducible to its effectiveness as an input-output function. But I agree the vatican's statement seems to go beyond that.
> Drawing an overly close equivalence between human intelligence and AI risks succumbing to a functionalist perspective, where people are valued based on the work they can perform
One might concede that AI can produce a good enough simulation of an embodied intelligence, while emphasizing that the value of human intelligence per se is not reducible to its effectiveness as an input-output function. But I agree the vatican's statement seems to go beyond that.