Intro to Interpretable AI - Lecture 1: The Where and Why of Interpretability

This class will focus on understanding the definition of interpretability. The variety in the objectives of early interpretability works, i.e. debugging, justifying outcomes, evaluating robustness or fairness, led to inconsistencies in how interpretable AI is defined. We will look at the definition and taxonomy from multiple perspectives, e.g. the legal, ethical and social perspective. Finally, we will analyze where interpretability is a requirement for the analyses, the trade-off between interpretability and complexity and the motivation for explaining AI decisions.