Fool Me Twice, Shame on Me: Being Deceived by a Robot Does Not Make People More Cautious

Abstract

Due to the distribution of various deceptive techniques in our daily environments, we must consider a way of educating interaction literacy, i.e., enabling trust calibration and appropriate responsive action through critical thinking and observations of interacting partners. For this purpose, in this study, we conducted an experiment where a social robot deliberately deceives participants under low-risk and non-harmful tasks (such as consent form readings) in both face-to-face and online survey settings, as an analogy of an evacuation drill or inoculation theory. We developed a semi-autonomous robot system where the robot provides information with or without deceptive strategies to attempt to either mislead (with deception) or encourage careful reading (without deception) of the consent form. After the tasks, the participants were informed of the robot’s role and evaluated their impressions of it. Additionally, following the laboratory experiment, participants voluntarily completed a follow-up online survey and participated in a similar consent form reading task online. The experimental results showed that the participants were more deceived due to the robot’s deceptive strategies, but their negative impressions of the robot were limited. Moreover, the participants who were deceived during the first task were also deceived in the second task. This suggests that our approach can be used as a promising method for identifying individuals who are easily deceived by others in non-harmful situations, thus aiding in the development of interaction literacy programs.