Interesting framing. The distinction between "tool does one thing when asked" vs "tool keeps going until goal is met" is actually solid. Quite interesting on the MIT study: only 5% of AI investments returning positive ROI even though everyone claims they're already using it.
That's very true. There's become a kind of "pressure" to use AI wether that is the best solution or not. Especially from places like admin/management, that thinks of it as this magical drug that solves everything.