Abstract
Machine learning has achieved significant success in many AI applications that have been deployed. Most early approaches focused on optimizing performance measurements such as accuracy. However, as machine learning techniques have been applied to fields that are highly sensitive to risk, such as healthcare, law enforcement, and finance, the trustworthiness of models, especially their explainability, has become an increasingly important concern. Fortunately, explainability has become a crucial aspect of various research directions. These research directions include explainable recommender systems, explainable natural language processing, explainable computer vision, explainable graph neural networks, and explainable fairness. In order to further improve the interpretability of machine learning models, some recent works in explainability have attempted to use causal reasoning techniques. In this chapter, we aim to provide an overview of causal explanation and discuss the design of Causal eXplainable Artificial Intelligence (CXAI).