The Bright and Dark Side of Consumer's Engagement with AI

Abstract

As AI-based digital voice assistants (DVA) such as Amazon's Alexa are becoming nearly ubiquitous, studies on consumer engagement (CE) with these voice assistants are emerging. However, this research field remains relatively nascent to-date. Drawing on the conceptual AI-based CE model proposed by Hollebeek and colleagues, this study examines AI-based CE nomological network (i.e., enablers, inhibitors, outcomes and moderating effects) using PLS-SEM. Based on data from 324 U.S. frequent users of DVA, the study validates the positive impact of DVA' usefulness, ease of use, anthropomorphism, social presence and social bonding on AI-based CE. Unexpectedly, the negative impact of privacy concerns and distrust was not supported. Further, social presence and anthropomorphism only positively impact AI-based CE when users mainly perform complex (vs. routine) goal-oriented tasks. Theoretically, the study adds to both CE and AI-based research by developing a first attempt to empirically validate Hollebeek and colleagues' conceptual model and the nomological network of AI-based CE. Moreover, we extend this model while exploring the moderating role of DVA task complexity. Managerially, the study contributes to a holistic understanding of AIbased CE and provides practical guidance on how to successfully manage such technologies.