Video Recordings

In this talk, we will start from the general motivation behind Explainable Artificial Intelligence (xAI) and point out the advantages of using natural language explanations (NLEs). However, NLEs, as well as other xAI methods, face difficulties in providing both plausible and faithful explanations, i.e. explanations that are trusted by users and reflect the real decision-making process of the AI model. We will illustrate this problem for two NLE methods for item recommendation and image classification tasks. In addition, we will explore simple NLE techniques for improving faithfulness while maintaining high plausibility.