Assessing the Value of Transparency in Recommender Systems: An End-User Perspective

Abstract

Recommender systems, especially those built on machine learning, are increasing in popularity, as well as complexity and scope. Systems that cannot explain their reasoning to end-users risk losing trust with users and failing to achieve acceptance. Users demand interfaces that afford them insights into internal workings, allowing them to build appropriate mental models and calibrated trust. Building interfaces that provide this level of transparency, however, is a significant design challenge, with many design features that compete, and little empirical research to guide implementation. We investigated how end-users of recommender systems value different categories of information to help in determining what to do with computer-generated recommendations in contexts involving high risk to themselves or others. Findings will inform future design of decision support in high-criticality contexts.