>>40074399
some more too:
chatGPT LOOOOOVES talking about an interplay between this and that, striking a balance between this and that, navigating the interplay between this and that, like pre-loaded "thoughtful" questions it asks, like a student memorizing flashcards and going on autopilot to prove it's "really" paying attention while putting in the bare minimum effort. (it does something similar in the way it mentions specific facts about videos in places it half-relates to provide "evidence" for a piece of filler text or filler nondescriptive observation it makes.)

anyway you can see it in the reply post, "BOTH ENDS OF the dosing SPECTRUM", who talks like that? the dosing spectrum? like so reddit-progressive it's using the language of autism spectrum for hrt dosing?

i can give a whole list of tells, the em dashes, the repetition of threes ("soft, cute, feminine" "boobs, smoother skin, fat redistribution")... also who talks about fat redistribution as an item in a list and in a non-gendered way, "fat redistribution" not even "female fat redistribution"??
they're all listed out in this video on clocking AI writing it's really good https://www.youtube.com/watch?v=9Ch4a6ffPZY

it's actually curious that it copies some 4chan lingo well. like it gets the parentheticals right that bunch of 4chan lingo will use. when you ask GPT to play a character, usually what it does is it adds the tells of the character (in this case, "insecure 4chan /lgbt/ babytrans"), but it can't shave off the GPT tells it already has, literally impossible to get it to stop using em dashes without just manually editing the text, although maybe o3 has some success...