Abstract
Generative AI chatbots are transforming e-healthcare by enabling scalable and efficient online consultations. However, there is limited understanding of how the design features of human-to-chatbot interactions shape user trust, a critical factor in technology acceptance. This study addresses the underexplored role of AI identity (AII)—the extent to which users perceive and relate to a chatbot as an autonomous agent—and its impact on reducing reliance on human consultations. Drawing on theories of identity disclosure, social presence, and AII, a between-subjects (2x2) experimental design was conducted with 327 participants interacting with a health consultation chatbot to examine the interplay between perceived trust, AII, and the need for human interaction. Results show that explicitly disclosing the chatbot’s AII significantly enhances trust. Moreover, AII fully mediates the relationship between trust and reduced preference for human consultations, indicating that stronger AII is associated with greater acceptance of AI-driven e-healthcare services. Interestingly, AI identity disclosure was found to boost trust, aligning with the uncanny valley hypothesis in human-AI interactions. These findings advance theoretical understanding of trust dynamics in human-AI systems and contribute substantially to the scholarly discourse on human-AI interaction by deepening insights into users' psychological orientations toward anthropomorphism and their preferences for delegating decision-making to AI-driven agents.