Explainable AI: A key driver for AI adoption, a mistaken concept, or a practically irrelevant feature?

Abstract

Explainable artificial intelligence (xAI) has become a popular subject of research amongst AI scholars in the last years. Some scholars consider xAI a significant driver of AI adoption in practice. However, at date, only a few studies investigated the conditions under which xAI solutions provide benefits in practice. Additionally, there is still a lot of controversy and inconsistency about related terminology revealing large conceptual differences between the understanding of explanations from a theoretical social science viewpoint and from a technological viewpoint. In this article, we strive to contribute to a more realistic picture of the potential and practical application scenarios of xAI. Thereby, we clarify the question whether xAI is a key driver for AI adoption, a mistaken concept from a theoretical point of view or perhaps a practically irrelevant feature and bridge the gap between different disciplines.