more specifically, I think an apology "generated with gen AI" fails not because it's written "by an algorithm" (i.e., a non-person entity who is not the apologizer), but because it's essentially co-written by the people who made the tool, whose *stated purpose* is to make it possible to generate things that look authentic but are not, using statistical prediction. this feels like splitting hairs, i guess, but i think it's important to trace that part of the authorship