really, if you are going to use the term "AI", even though it's a per se deceptive marketing phrase with a racist backstory, you should distinguish it from ML, which is a collection of mature technologies that are often useful and efficient and embedded in all kinds of technology (e.g. phones, cars)
the Merlin Bird ID app uses a deep-learning algorithm trained exclusively on sound samples from the Macaulay Library of Bird Sounds at Cornell University. This is very different, both in scope & impact, from ChatGPT & other LLMs that have been unleashed to scrape content from the entire internet.
if you look at the titles of papers from the big conferences in the space people do not talk about "AI" because it is, as a technical term, meaningless.
there's not a bright line, necessarily, technically, but if you want a high level rubric for when somebody is doing something pernicious and stupid with these (casting a wide net) technologies, somebody who is nominally an expert who speaks uncritically about "AI" or (god help me) "AGI" isn't bad
I wrote about this in more detail, see below, but the definition of “intelligence” the early AI researchers were working from has been shown robustly to be racially biased and nobody doing “AI” these days has convincingly advanced a better one.
