Unlocking Privacy in Healthcare: The Impact of Explanations on Privacy Concerns and Self-Disclosure to Conversational Technologies

Abstract

While artificial intelligence (AI)-based conversational technologies offer exciting prospects in healthcare, the lack of transparency and elevated privacy concerns in using such technologies remain a challenge and make much-needed information difficult to obtain while administering patient care. Approaches that emphasize transparency and interpretability of AI systems provide a promising avenue to address these concerns. In this study, we explore the role of transparency-enhancing explanations as a way for caregivers to elicit truthful disclosure of otherwise private information from patients. Specifically, we explore how automated explanations provisioned by conversational technologies can help reduce the user's privacy concerns and bring about self-disclosure, thus helping to improve key outcomes such as accurate diagnosis and effective treatment. Through an online experiment with 556 participants in a healthcare context, we uncover the mediating effects of two critical factors, informational justice and perceived relevance, on privacy concerns. We find that explanations foster perceptions of informational justice and perceived relevance in the user, which help reduce privacy concerns and bring about self-disclosure. The study's findings have implications for researchers as well as practitioners who leverage conversational technologies in healthcare and other service contexts.