if you look at arguments in favor of genAI you won't find any

but you will find a lot of assertions in favor of genAI

because that's how you have to talk to an autocomplete algorithm in order to try to make it give you the words you want

you can't convince an autocomplete algorithm that something is good for a reason because autocomplete algorithms don't think

they just guess the most likely next word using (frankly less complicated than what you've probably already done today, just on a ridiculous amount of numbers) math

so if you put more stuff in the document that's being autocompleted about how the thing is good, then the guessed words will reflect that