A primer on interpretability and how AI researchers are figuring out how to open and understand the "black box" that holds the formulas within most AI models (Oliver Whang/New York Times)