If we don’t understand how machine learning works, how can we trust it? Increasing model transparency creates risks as well as rewards.
