If so, what are its intentions?
Fundamental to our human use of language, like all of our actions, are our intentions: Why we say and write what we do to our listeners and readers. We attempt to modify their own cognition and behavior in some ways — by informing, questioning, requesting, commanding, persuading, entertaining, confusing, deterring them, and so on.
To the extent that systems like ChatGPT lack models of their human readers and intentions with respect to those readers, they are deficient as cognitive agents. Perhaps one can say that their designers have built them with an implicit intention of informing their readers, but like some people, what they “know” is mostly based on just what they themselves have read. The best that one can say for them in that regard is that they are more “widely read” than most people.
So their structure is not likely to be a good model for how our human brains create our language behavior.
Have I persuaded you?
