Fundamentals on explainable and interpretable artificial intelligence models

The chapter provides the notions and tools to illustrate what explainability and interpretability imply in the context of AI-based applications, what is required to develop explainable AI (XAI) methods and how this can be assessed. The difference between the concepts of interpretability and explainability is outlined and two main approaches for XAI are discussed, namely interpretability by-design and post-hoc explainability. To complement theoretical information with practical/applicative insights, an example of adoption of XAI approaches is showcased for a healthcare-related case study.