Interpretability

Interpretability seeks to understand how a model processes information and reaches outputs, using analyses of behavior, representations, parameters, or internal computations.

Also known as Explainable AI, XAI.