zlacker

What hypothesis of Chomsky are you guys talking about? If it is about innateness of grammar in humans then obviously this can not be shown wrong by LLMs trained on a huge amount text.

>>uecker+(OP)
Chomsky's claim is that linguistics is actually a branch of cognitive science, that language is, by definition, "what the brain does" and that "meaning" in language is grounded in the brain, by what the speaker as a biological entity intends.

But this forces one into the position that whatever a LLM is doing is not real language, just an imitation of language.

If you take the fact that LLMs are emitting "real language" at face value, then you need to adopt a more structuralist view of language, in which "meaning" is part of the system of language itself and does not need to be grounded biologically.

I don't think holding a structuralist view of language precludes believing that humans have a biological facility for language, or even that language is shaped and ultimately a result of that biological facility. Its more an argument over what language IS -- a symbolic system, or an extension of the human brain.

>>lukev+Wac
Thank you for the explanation. I do not quite understand this though and I am also not a Chomsky expert. But I do not understand why "this forces one into the position that whatever a LLM is doing ... is just an imitation of language" or why LLMs somehow contradict the idea that "language is what the brain does" (and we already know that language is by far not the only thing the brain does). I am curious why you think that Chomsky's position is that "meaning of the language has to be grounded biologically." Does he say this explicitly somewhere or is this your (or someone's else) interpretation of what he says? I am asking because this is not matching what I read about Chomsky's ideas (but I did not read much about it, so this may just be my ignorance).

>>uecker+5Fd
Putting Chomsky aside, LLM do not seem to teach us much. Obviously, they are good at language processing and for just being trained on language they seem to appear surprisingly "smart". LLMs after training also have language built-in similar to what the human brain has after evolution, and this seems to confirm rather to contradict the idea of language being an important and fundamental part of the the human brain does. On the other hand, there is also a lot missing. But we already know that there are many other cognitive processes in the human brain which are not related to language, so it is not something we learned recently by analyzing deficiencies of LLMs.

If you believe in esoteric stuff such as "qualia", then LLMs also tell you nothing. You can continue to believe that true experience requires a human brain and all that a computer does is imitation. But this has no observable consequences that can be used to falsify these ideas, so is not a scientific concept in the first place.