Ethical tensions in AI-mediated clinical communication: a mixed-methods study of nursing students’ perspectives

Abstract

The increasing use of artificial intelligence (AI), particularly large language models (LLMs), is reshaping clinical communication by influencing how patients access, interpret, and act upon clinical information. Recent conceptual work has suggested that AI-mediated communication raises ethical questions concerning interpretive authority, professional responsibility, and the communication of clinical uncertainty. However, empirical evidence regarding how future healthcare professionals understand and respond to these emerging ethical challenges remains limited. Guided by a conceptual framework comprising the ethical dimensions of interpretive authority (who interprets and legitimises clinical meaning), professional responsibility (who remains accountable for patient understanding), and the communication of clinical uncertainty (how uncertainty is explained and negotiated), this study explored perceptions of AI-mediated clinical communication. An exploratory convergent mixed-methods design was employed. Thirty-two undergraduate nursing students (16 international and 16 local) participated in a theory-informed case-based session involving patients using AI-generated health information during communication across a surgical care pathway. Qualitative written responses were analysed using Braun and Clarke’s reflexive thematic analysis approach (2022). Following thematic development, a separate structured thematic quantification procedure was undertaken as a mixed-methods strategy to facilitate descriptive comparison across participant groups rather than to modify the qualitative analysis itself. Given the exploratory nature of the study, analyses focused on identifying patterns of perception and interpretation rather than causal effects. Descriptive quantitative findings showed increased mean scores across all three ethical domains following the case-based session, with the greatest increases observed in interpretive authority and the communication of clinical uncertainty. Qualitative analysis identified three overarching dimensions: perceptions of AI in clinical contexts, ethical and communication tensions in AI-mediated care, and professional roles in AI-mediated communication. Participants recognised challenges associated with patients’ over-reliance on AI, difficulties in correcting AI-influenced patient understanding, and tensions arising from the communication of uncertainty. They also described an expanded professional role involving interpretive mediation, trust-building, and contextualisation of AI-generated information. The empirical findings indicated that participants perceived these issues as important ethical challenges in responding to patients’ AI-informed understanding, although variation was observed in the relative prominence of individual themes across participant groups. The findings suggest that AI-mediated clinical communication introduces identifiable ethical tensions concerning authority, responsibility, and uncertainty. Within the educational scenarios examined, participants recognised that AI-generated explanations accessed by patients may reshape how clinical meaning is interpreted and negotiated while leaving healthcare professionals’ responsibility for supporting patient understanding largely unchanged. By providing empirical evidence informed by a conceptual framework of AI-mediated communication, this study contributes to emerging discussions regarding the ethical implications of AI in healthcare communication and highlights the importance of addressing authority, accountability, and uncertainty within the responsible governance of AI-enabled healthcare communication.