Revealing While Solving: Exploring the Strategy of Service-Oriented Self-Disclosure of AI Chatbots

The self-disclosure of AI chatbots has become an important design mechanism for improving transparency and social engagement during ongoing interactions. This study proposes a service-oriented framework for implementing AI chatbots’ self-disclosure and conducted a 2 (self-disclosure content: relation-oriented vs. transparency-oriented) × 2 (self-disclosure delivery style: proactive vs. on-demand) × 2 (interaction context: social and instrumental) experiment using chatbots developed with the DeepSeek API, to investigate how AI chatbots’ self-disclosure strategies influence users’ experience across different interaction contexts. The results show that in social interactions, users significantly preferred AI chatbots using proactive delivery style and expressed higher continuous usage intention for AI chatbots with relation-oriented self-disclosure content. No significant preference emerged in instrumental interactions. A significant gender effect was also observed, with male users showing greater sensitivity to relational cues delivered by AI chatbots. In addition, a fundamental principle of human–AI self-disclosure was found: users not only appreciate emotional responsiveness but also expect practical, problem-solving support.