This was an excellent overview article. As a data scientist with a degree in artificial intelligence, I am familiar with the idea that “real" AI is always a point over the horizon, and everything up to now is just “clever programming”. There have been laws passed in Europe about machine learning that stipulate that certain decisions made by an AI (to deny benefits of various sorts, mainly) must be able to fully explain their reasoning to a human, or those decisions are not valid. I always thought that was a fun part of ML…trying to reverse engineer the meaning of certain nodes in a neural network.