Why Computer Scientists Can’t Explain How Their Machine Learning Algorithms Actually Work A pretty good high-level explanation of the black-box problem found in e.g. neural networks.