Excellent discussion James. But AI is far move than a semantic problem. What if we called it Algorithmically Arbitrary Gatekeeper (which is what it is). One's entire life can be red-lined based upon past dicey zip code (which when combined with a 1000 irrelevant amplified factors), via a cohort aggregation episode of machine learning has determined they are a highly probable high risk individual. I just read a piece in the NYT about how banks are dumping (and sending into chaos) small businesses and no human can even identify what they did wrong. What happens when augmenting really does mean replacing? Or worse, what happens when economists sub machina economicus as the metric of rational actor. Note:

https://medium.com/@daneke/automacene-rising-the-perilous-political-economy-of-artificial-intelligence-75aefd7f1979