The author has written about systems that defy detailed understanding as a follow up to this article in which he mentions Deep Reinforcement Learning Models as an example. It's a good read, especially in combination with this one.