This talk will overview recent research on human tendency to over-trust and under-trust AI. It will describe the role played by cognitive heuristics (or mental shortcuts) in shaping user perceptions, with a particular focus on machine heuristic. With data from experiments comparing human and AI sources and gatekeepers, it will delineate the effect of both positive and negative machine heuristics upon user trust. Finally, it will suggest ways to mitigate the effect of heuristics and achieve better trust calibration.
S. Shyam Sundar | Donald P. Bellisario College of Communications