Slide to Explore 'What If': An Analysis of Explainable Interfaces

Most users interact with explainable AI (XAI) systems through an interface. Despite the growing assemblage of human-centered XAI literature, there is still limited research that investigates how explanations are presented in interfaces. With this preliminary study, we propose a novel approach to analyze the content and interactions of explainable interfaces by mapping UI elements and interactions to the type of explanation they convey. Based on analysis of 28 explainable interfaces, we offer initial findings of how various interaction concepts and UI elements are used to provide different explanations.