Health Inquiry with AI: How Empathetic Expression and Conversational Contexts Shape Users' Communicative Acts

As online health information-seeking shifts to conversational AI, high-quality information retrieval increasingly relies on users’ “communicative acts”(proactively sharing and seeking information)— similar to how effective diagnosis and personalized guidance are elicited in patient-clinician communication. Drawing on health communication research, this study examines how a chatbot’s modality of empathetic expression (Verbal, Visual, Multimodal) and the conversational context (General, Sensitive, Mental Health) influence these acts through a 2 × 2 × 3 within-subjects experiment (N= 48). The results revealed that while verbal and multimodal empathy significantly increased reply length, communicative acts were largely shaped by conversational context, with Sensitive context triggering more question-asking and Mental Health context leading to heightened concerns, assertive responses, and unprompted information disclosure. Combined with qualitative findings, we discuss design implications for building context-sensitive AI health inquiry systems that can encourage active user participation.

Introduction. Driven by the rapid evolution of large language models (LLMs), online information-seeking, particularly health inquires, is shifting away from traditional search engines to conversational AI [1, 35, 39]. Yet, the quality of information retrieved from AI depends not only on how well the model is designed and trained [10, 28], but also individual’s ability to communicate with (or prompt) the AI, such as asking follow-up questions or adding contextual details about their health conditions [13, 15, 18, 22]. This is similar to effective communication with human clinicians, where the patient’s active inquiry and disclosure are essential for the clinician to fully understand their needs and concerns, and deliver more appropriate, personalized guidance [4, 6, 31]. Health communication researchers have characterized this active participation as “communicative acts,” which include asking questions, expressing concerns, and making assertive responses [31]. These acts allow patients to articulate their needs, worries, preferences, and interpretations during clinical encounters [31].

Discussion / Conclusion. Our analysis showed that while the modality of empathetic expression increased participants reply length, such increase did not necessarily contribute to communicative acts. This finding suggested that while empathetic cues (particularly verbal and multimodal ones) are effective at keeping users engaged and encouraging them to type more, the additional text may consist of conversational politeness or social reciprocity [3, 8] rather than proactive information-seeking or assertive behaviors. On the other hand, we found that participants’ communicative acts were mainly driven by the conversational context—the complexity and urgency of their health conditions. This indicates that active participation, such as intensive information-seeking, sharing one’s diagnostic interpretations, and volunteering unprompted context, is fundamentally a problem-driven coping mechanism rather than a reaction to the AI’s social behaviors [16, 21, 30].