(There *is* a problem where "AI" is an incredibly elastic word, whose current meaning is mostly shaped by marketing. I personally would, if we were in some sort of perfectly abstract Athenian Forum philosophical debate, distinguish between "machine learning" and post-2018 LLM/genAI, with the gist being that machine learning is sometimes effective [though usually not optimal] for detection and classification, but cannot *generate* text or images for a function other than pure aesthetics.)