Exploring the Potential of Ambient AI for Inpatient Documentation: A Qualitative Study with Junior Doctors.

Clinical documentation is essential for safe and effective patient care but places a substantial clerical burden on doctors, particularly those early in training. Ambient artificial intelligence (AI) systems, which passively capture clinical conversations and generate structured notes, have demonstrated promise in outpatient and primary care settings. However, their use in inpatient ward rounds remains largely unexplored. Semi-structured interviews were conducted with ten postgraduate year one doctors following participation in simulated orthopaedic ward rounds incorporating an ambient AI scribe (Heidi Health, Melbourne, Australia). The topic guide was informed by Normalisation Process Theory (NPT) to explore participants' understanding of AI-assisted documentation workflows, the work they anticipated would be required at individual, team, and system levels for implementation, and their appraisal of the potential benefits, risks, and practical challenges of future use. Data were analysed collaboratively by two researchers using reflexive thematic analysis with a hybrid inductive-deductive approach. Seven themes were identified and mapped across the four NPT constructs. Coherence was reflected in participants' understanding of ambient AI as a way to reduce documentation burden, support patient-centred records, and improve accessibility. Cognitive participation was evident in their recognition that implementation would require individual engagement, including review and approval of AI-generated outputs. Collective action captured the practical work required to adapt ward round behaviours, redistribute documentation tasks, manage consent, and address infrastructure, access, and governance requirements. Reflexive monitoring was demonstrated through participants' appraisal of anticipated benefits, including potential for improved clinical engagement, multidisciplinary communication, and real-time capture of complex conversations, balanced against risks of clinical misrepresentation, cultural resistance, and privacy concerns. Through simulated exposure, participants made sense of ambient AI scribes as a potentially valuable addition to inpatient documentation workflows, anticipating effects on efficiency, communication, and patient-centred documentation. However, they also identified individual, team, and organisational work required for implementation, including workflow adaptation, consent processes, data privacy, access controls, infrastructure, and real-world appraisal in clinical settings.