AI-generated health communication to understand how people interpret client narratives: Interpretive variability and its implications for trust and biopsychosocial care prioritisation

Little is known about how health communication and client narratives are interpreted across different user groups. This study examines how AI-generated videos, used to emulate client health communication, are interpreted and evaluated, with a focus on variability in care prioritisation and perceptions of AI-mediated communication. A cross-sectional online survey incorporating AI-generated, one-way video case studies was conducted (n=276; 93 health professionals and 183 members of the public). Participants were randomly assigned to one of four AI-generated video scenarios featuring a standardised core client narrative with variations in sociocultural context. Participants ranked care priorities, indicated preferences across lifestyle factors, and evaluated perceived impacts of AI-mediated communication. Across both groups, psychological and lifestyle support were consistently prioritised over medical interventions, reflecting a shared biopsychosocial orientation to care. However, interpretation was not uniform. Members of the public demonstrated greater variability in prioritisation across the different videos, suggesting sensitivity to sociocultural cues despite consistent clinical content, whereas health professionals showed more consistent response patterns. Perceptions of AI-mediated communication were characterised by ambivalence rather than rejection, with concerns centred on loss of human touch, potential miscommunication, and the need for human oversight. These findings indicate that the interpretation of standardised AI-generated videos of client narratives is not consistent across users. Instead, responses vary according to user group and contextual cues. Digital mental health systems should therefore be designed to support interpretive variability, preserve human oversight, and maintain relational aspects of care in sensitive clinical contexts.