Interpretability

The degree to which a human can understand how a model arrives at its output.

Large models are often hard to interpret, motivating explainability tools.