As it stands, building artificially intelligent machines that can teach themselves is not only really difficult, it’s a painfully slow process. Looking to accelerate things, next month DARPA will begin to crack the whip with a program called Probabilistic Programming for Advanced Machine Learning (PPAML). Besides seeking to dramatically increase the number of people who can successfully build machine-learning applications, the program wants to radically increase the effectiveness of machine learning experts so they can create new applications beyond the limits of technology that’s currently available. “Our goal is that future machine learning projects won’t require people to know everything...