Users’ perceptions of voice assistants’ effectiveness: the interplay of trust, intelligence, information accuracy, and usefulness

Voice Assistants (VAs) are AI-powered programs that use Machine Learning, voice recognition, and natural language processing to answer user questions and perform various tasks. Despite the abundant studies of VAs, scarce research has examined the factors users employ in evaluating the effectiveness of their VAs. This study examined college students’ VAs’ usage, perceptions of trust, intelligence, usefulness, and information accuracy. It investigated the interplay between and among the participants’ perceptions of these factors in their VAs. This study's findings revealed that while most participants highly trusted their VAs' information accuracy, their conceptions of how they functioned were lacking. One-third of the participants’ VAs failed to provide accurate information, resulting in frustration and confusion. Although we found an interplay between and among users’ perceived VAs’ information accuracy, trust, intelligence, and usefulness, a significant correlation existed between users’ perceived trust and the information accuracy it provided. VAs’ accent influenced the participants’ trust in information accuracy. This study’s findings have implications for designing user-centered VAs, improving users’ interactions, and empowering users with AI literacy skills.