Abstract
This study examines the application of AI voice assistants in Chinese postpartum follow-up phone calls, with particular focus on how interaction design strategies influence users' self-disclosure intention. A 2 (voice gender: female/male) × 3 (self-disclosure strategies: normal conversation without additional disclosure/objective factual disclosure/emotional and opinion-based disclosure) mixed experimental design (n = 395) was conducted to analyze how the gender and self-disclosure strategies of voice assistants affect users' stereotypes (perceived warmth and competence), and how these stereotypes, mediated by privacy calculus dimensions (perceived risks and perceived benefits), influence self-disclosure intention. The experiment measured various indicators using a 7-point Likert scale and performed data analysis through analysis of variance (ANOVA) and structural equation modeling (SEM). The results demonstrate that female voice assistants significantly enhance users' perceived warmth and competence, while emotional self-disclosure strategies significantly improve perceived warmth. Stereotypes about the voice assistant positively affect users' self-disclosure intention through the mediating effects of perceived risk and benefit, with perceived benefit exerting a stronger effect than perceived risk. These findings provide valuable insights for the design and application of AI voice assistants in healthcare, offering actionable guidance for enhancing user interaction and promoting self-disclosure in medical contexts.