LLMs are simply tools that emulate the communicative function of language, not the separate and distinct cognitive process of thinking and reasoning …
Take away our ability to speak, and we can still think, reason, form beliefs, fall in love, and move about the world; our range of what we can experience and think about remains vast.
But take away language from a large language model, and you are left with literally nothing at all.
The author seems to be making the assumption that a LLM is the equivalent of the language processing parts of the brain (which according to the cited research supposedly focus on language specifically and the other parts of the brain do reasoning) but that isn’t really how it works. LLMs have to internally model more than just the structure of language because text contains information that isn’t just about the structure of language. The existence of Multimodal models makes this kind of obvious; they train on more input types than just text, whatever it’s doing internally is more abstract than only being about language.
Not to say the research on the human brain they’re talking about is wrong, it’s just that the way they are trying to tie it in to AI doesn’t make any sense.
Took a lot of scrolling to find an intelligent comment on the article about how outputting words isn’t necessarily intelligence.
Appreciate you doing the good work I’m too exhausted with Lemmy to do.
(And for those that want more research in line with what the user above is talking about, I strongly encourage checking out the Othello-GPT line of research and replication, starting with this write-up from the original study authors here.)
The author seems to be making the assumption that a LLM is the equivalent of the language processing parts of the brain (which according to the cited research supposedly focus on language specifically and the other parts of the brain do reasoning) but that isn’t really how it works. LLMs have to internally model more than just the structure of language because text contains information that isn’t just about the structure of language. The existence of Multimodal models makes this kind of obvious; they train on more input types than just text, whatever it’s doing internally is more abstract than only being about language.
Not to say the research on the human brain they’re talking about is wrong, it’s just that the way they are trying to tie it in to AI doesn’t make any sense.
Took a lot of scrolling to find an intelligent comment on the article about how outputting words isn’t necessarily intelligence.
Appreciate you doing the good work I’m too exhausted with Lemmy to do.
(And for those that want more research in line with what the user above is talking about, I strongly encourage checking out the Othello-GPT line of research and replication, starting with this write-up from the original study authors here.)