What are the two major approaches to model interpretability in Explainable AI (XAI) as discussed in the chapter?

The two major approaches to model interpretability in Explainable AI (XAI) are post hoc explainability and intrinsic explainability. Post hoc explainability involves techniques applied after a model is trained to interpret its predictions, while intrinsic explainability refers to models designed to be inherently interpretable from the outset.

Post hoc explainability includes methods that provide insights into model predictions after the model has been trained, allowing for interpretation without altering the model itself. This approach is model agnostic and can be applied to various types of models. In contrast, intrinsic explainability involves creating models that are inherently interpretable, meaning their decision-making processes are understandable without additional explanation tools. Examples of intrinsic models include linear models and decision trees, which are designed to be simple and transparent.