Without interpretability, AI and machine learning models are black boxes that are hard for their users to apply, understand, and trust.