Leveraging Self-Disclosure and Utility Theory for Extracting Different Types of User Information with Conversational Agents

Leveraging Self-Disclosure and Utility Theory for Extracting Different Types of User Information with Conversational Agents

The more precisely an AI system collects and analyzes user information, the more effectively it can tailor future recommendations for each user. However, gathering comprehensive information for individual users remains a significant challenge because they may have concerns about privacy or find the process bothersome. To encourage users to willingly provide diverse and meaningful personal information, we applied two widely discussed concepts in psychology and economics to conversational agents: Self-Disclosure and Utility Theory. Our study revealed that while both conversational strategies influenced user experience, the Utility Theory strategy, when combined with questions targeting opinions and emotions, enhanced users' willingness to disclose personal information and improved their overall disclosure experience. These results highlight the importance of tailoring conversational strategies to information types to encourage self-disclosure effectively. Based on these findings, we propose design considerations for efficiently gathering user information through conversation.