Expectation vs Reality in Users' Willingness to Delegate to Digital Assistants

Abstract

We report results of a survey with 2500 US-based users of popular digital assistants (DAs) to understand and prioritize the factors that drive consumer adoption. Using structural equation modelling, we investigated the relationship between respondents' behavioural intentions to delegate tasks to their DAs and two predictor layers: DA (users' attitudes and familiarity with them) and task factors (need for control/transparency, subjectivity, risk, self-efficacy, and frequency), mediated by a values layer (trust, perceived ease of use and usefulness). Perceived usefulness and trust were strong direct predictors of willingness to delegate. Both DA-related factors had indirect effects, with a surprising negative influence of familiarity on both trust and usefulness. Our novel findings of task-related effects on the value factors may explain this disparity. These results imply a mismatch between users' expectations for DAs and their actual experiences. We interpret these findings in light of related work and derive implications for practitioners.