How should voice assistants be heard? the mitigating effect of verbal and vocal warmth in voice assistant service failure

Voice assistants have become increasingly popular touchpoints inAI-infused service encounters in the hospitality industry. Althoughwe have seen a growing body of research in this area, littleattention has been paid to specific service failures involvingexclusively voice interactions. Drawing from the Computers AsSocial Actors (CASA) research paradigm and the StereotypeContent Model, this research explores how warmth can mitigatethe negative consequences of service failure involving voiceassistants. In two experiments using both physiological (EDA) andpsychological measures, we show that the perception of warmthimproves consumers’emotional reactions and increases re-patronage intention following a negative service outcome. Wealso found that the optimal voice to be used for a voice assistantis a dynamic speech style combined with emotionally expressiveand warm verbal content. Thesefindings contribute to theknowledge of voice-based smart service interaction and provideinsight into how to mitigate the negative consequences ofservice failure involving voice assistants.