Sure it does. It perfectly deliniates it. LLMs are not: sensitive to causal structure, dynamically adapting to environmental changes, growing, developing sensory-motor capacities, they are not with us in our environment, they are not: expressing desires, preferences, intentions, beliefs, motivations, etc. And so on.
To say, "they just predict the next word" is literally to say that all apparent functions of an LLM are engineering tricks, circumstantially useful -- to be found by (largely software) engineers in building apps.
The reason any reply is given to any prompt is that this reply is maximally probabilistically consistent with a historical corpus of text.
This excludes the possibility the reply is say, an expression of a history of aesthetic experiences which form an individual's taste. Or, likewise, anything.
This is a scientific claim about what LLMs are, not an engineering claim about what over-hyped apps might be abled to do with them.
To say, "they just predict the next word" is literally to say that all apparent functions of an LLM are engineering tricks, circumstantially useful -- to be found by (largely software) engineers in building apps.
The reason any reply is given to any prompt is that this reply is maximally probabilistically consistent with a historical corpus of text.
This excludes the possibility the reply is say, an expression of a history of aesthetic experiences which form an individual's taste. Or, likewise, anything.
This is a scientific claim about what LLMs are, not an engineering claim about what over-hyped apps might be abled to do with them.