This, via @emilymbender, is so good: a paper opposing " gratuitous anthropomorphic features." I've been arguing that LLMs should not use first-person human but first-person machine and not use brain verbs but machine verbs (i.e., instead of "I write," "the program assembled).
Follow
@jeffjarvis @emilymbender Yeah, but how do you do that? You'd have to convert all the input data it learns from to use that grammar, and then it would start quoting people and articles as if they were written by a computer.
Properly thinking machines should be able to be taught to do what you describe. But these mad libs LLMs are only as good as their input.
