Prof. YU Wenting, Department of Language Science and Technology

The Impact of AI's Self-Disclosure of Limitations on Human-AI Trust and User Experience. Symposium for Young Scholars on Human-Machine Communication. Huazhong University of Science and Technology, Wuhan, 16 May 2026.

Abstract
This study examines how an AI agent’s proactive disclosure of its own limitations affects users’ trust and experience in human–AI interaction. Participants interacted with an AI avatar that either disclosed its limitations in competence, or provided no such disclosure. The results show that flaw self-disclosure did not significantly affect social presence, perceived competence, or liking of the AI avatar. However, it significantly enhanced users’ benevolence-based trust, suggesting that interpersonal communication patterns do not apply to human– AI interaction communication all the time. The findings responded to Media Are Social Actors (MASA) framework and provided empirical evidence for human-media scripts.