Abstract
In the era of the Internet of Things (IoT), emerging artificial intelligence (AI) technologies provide various artificial autonomy features that allow intelligent personal assistants (IPAs) to assist users in managing the dynamically expanding applications, devices, and services in their daily lives. However, limited academic research has been done to validate empirically artificial autonomy and its downstream consequences on human behavior. This study investigates the role of artificial autonomy by dividing it into three types of autonomy in terms of task primitives, namely, sensing, thought, and action autonomy. Drawing on mind perception theory, the authors hypothesize that the two fundamental dimensions of humanlike perceptions—competence and warmth—of non-human entities could explain the mechanism between artificial autonomy and IPA usage. Our results reveal that the comparative effects of competence and warmth perception exist when artificial autonomy contributes to users' continuance usage intention. Theoretically, this study increases our understanding of AI-enabled artificial autonomy in information systems research. These findings also provide insightful suggestions for practitioners regarding AI artifacts design.