Explanations of AI Are Greater than Explainable AI
The eXplainable AI (XAI) community has traditionally focused on algorithm-centric methods to facilitate understanding. However, insights from related fields suggest alternative strategies for making AI systems understandable to users. In this positional paper, we argue that methods used in human-robot interaction (HRI) and human-computer interaction (HCI) can effectively support user understanding—often without relying on traditional XAI techniques, especially for end-users with limited AI expertise. Examples include user onboarding and communicating system states. While XAI methods can help generate such explanations, they are not always necessary, as users have diverse informational needs. We argue that explanations of AI extend beyond what is usually considered by the XAI community and that the XAI researchers would benefit from looking at how closely related fields make complex…