Indeed, this article could easily - and possibly should have - equate imitators to current AI (which tend to fail on deep knowledge, and - unlike experts - show inflexibility when challenged).

Which is not to say that shallow knowledge (whether AI or human) is a bad thing. I'm very happy to be helped out by ChatGPT in places where I'd be in the shallow knowledge ('imitator') camp. Still, the limitations of current AI seem quite similar to the 'imitators' described here (noting that plenty of humans aren't above making stuff up either).