Trust in Social Robots: Experimental Outcomes and Interactional Trajectories

Abstract

This study examines interactions between students and a social robot in an experiment designed to explore students' trust in educational robots. In the experiment, 16 out of 22 students come to accept an erroneous solution to a mathematical task presented by the robot, even though the vast majority of students initially suggest the correct solution. While the experiment initially yields two outcomes—acceptance (16 students) and rejection (six students)—our analyses demonstrate that there is significant variation among students in how they approach the interaction and the robot's advice, even within each outcome. Drawing on an ethnomethodological and conversation‐analytic (EMCA) perspective on trust, we argue that the precise ways in which students accept or reject the robot's solution must be seen in terms of their understanding of the educational and experimental context; that is, their definition of the situation in terms of what kind of activity is going on, and the role of the robot within it. The findings highlight the need for an interactionally sensitive approach to interpreting experimental studies of trust in human–robot interaction.