Why overtrust in social robots is inevitable
TL;DR
In this paper, experimental evidence is presented that it is possible to block overtrust to some extent, but that regulating people’s trust in social robots remains challenging.
Abstract
When robots enter our social spaces, we may view them from three different perspectives: as machines, as artefacts with specific affordances (like ‘seeing’, ‘speaking’, ‘dancing’), or as nonstandard characters, which people construct based on the features of the robot depicted, the way it is framed by others, and influenced by their own needs and dispositions. People can switch seamlessly between the three perspectives, for example, offering polite assistance in one minute and turning it off for cleaning in the next. However, when robots are construed as characters, they become social actors, and with this sociality, also expectations concerning trust apply. The fact that they participate in activities that rest on human social interaction mechanisms is likely to lead to overtrust in these characters since it evokes general pragmatic principles, like recipient design, grounding and drawing on common ground. In this paper, I present experimental evidence that it is possible to block overtrust to some extent, but that regulating people’s trust in social robots remains challenging.