Evoking trust in smart voice assistants

Purpose To unravel the drivers of service consumers’ parasocial relationships with artificial intelligence-enabled voice assistants (VAs), this study examines how VA frequency- and time-related paralinguistic features affect parasocial attraction of VAs. The authors zoom in on the interrelations between consumers’ social perceptions by exploring how parasocial attraction drives perceived anthropomorphism and trust in VAs. Design/methodology/approach In an online experiment, a VA displayed high or low voice intonation and high or low speech rate. Self-reported data of 580 Prolific participants regarding their perceptions of parasocial attraction, anthropomorphism and trust were collected and subjected to partial least squares path modeling. Findings The results show a moderating role of VA speech rate on the effect of voice intonation on parasocial attraction, such that voice intonation increases VA parasocial attraction when speech rate is high. In turn, parasocial attraction drives trust in a VA, both directly and indirectly via perceived anthropomorphism. Research limitations/implications The study outcomes can help designers and service managers design and infuse VAs in service frontlines within smart service systems, in ways that promise to enhance customer experiences and make services more inclusive. Originality/value By addressing the interplay between VAs’ frequency- and time-related paralinguistic features, this study offers new insights into the effects on consumers’ parasocial relationships with VAs and subsequent social perceptions. Such insights can benefit continued research into smart service systems.