Understanding the minds of voice assistants—How do perceived technical understanding and mind attribution influence trust and satisfaction over time?

Voice assistants are widely used, but like other AI-based dialogue systems, their functioning is opaque to users. Little is known about how the (mis)understanding of the system’s human-like functioning influences users’ trust. This study scrutinizes how the (perceived) understanding of a voice assistant’s functioning and the attribution of a (human-like) mind influence trust in the system and the resulting satisfaction. The study employed a six-wave repeated-measures online survey (N = 128), which was conducted over the course of 2.5 years. Results indicate an increase of trust over time, which develops despite decreasing usage frequency. Perceived technical understanding and (mis)interpretation as human-like emerge as predictors of trust but not of satisfaction. Noteworthily, perceived technical understanding does not increase with long-term usage while the importance of understanding the device was even rated lower over time. Also, perceived technical understanding and perceived importance of understanding did not preclude a human-like mind (mis)attribution.