Does Excessive Autonomy Undermine Trust? The Influence of AI Agent Autonomy on User Trust and Continuance Usage Intention

• 1、School of Management,Wuhan University of Science and Technology,Wuhan 430065

Abstract：Against the backdrop of exercise and health management, this study examines the underlying mechanism through which AI agent autonomy affects user trust and continuance usage intention. A nonlinear model was developed and tested to investigate the effect of AI agent autonomy on user trust based on the theories of human-machine trust, value theory, and the theory of perceived control. These research findings reveal an inverted U-shaped relationship between AI agent autonomy and user trust, with a trust threshold. Perceived value plays a curvilinear mediating role in this inverted U-shaped relationship, while perceived control plays a negative linear mediating role between AI agent autonomy and trust. Additionally, privacy risk perception strengthens the inverted U-shaped impact of AI agent autonomy on perceived value while amplifying its negative effect on perceived control. In the context of exercise and health management, this study demonstrates the dual mediation process by which AI agent autonomy influences user trust. It offers a theoretical foundation for optimizing autonomy levels and privacy communication tactics of health-related applications, as well as promoting long-term user retention.

Keywords： AI agent autonomy perceived value perceived control human-machine trust