AI Documentation Systems Quietly Shift Hospital Work from Staff to Doctors
A decade-long study of AI-powered clinical note-taking reveals an unintended consequence: hospitals are offloading quality control work from administrative staff onto busy clinicians. The finding challenges assumptions that AI automation reduces labor burden, suggesting organizations need to rethink how they implement these systems to avoid burnout and error.
Originaltitel: When AI enters clinical documentation: informal learning in health-care work
<p><strong>Purpose:</strong> The present research paper aims to examine how automated clinical documentation reshapes work practices and learning in a hospital setting. Focusing on automatic speech recognition (ASR), framed as artificial intelligence (AI), this paper explores how clinicians make such systems workable in practice and how this process brings organizational intelligence to the foreground. By examining a system used in clinical practice for over a decade, the study reveals how work practices and learning evolve around a technology that never fully stabilizes.</p><p><strong>Design/methodology/approach:</strong> The study is based on ethnographic fieldwork conducted between 2020 and 2022 in an orthopedic department at a large Danish public hospital. Our reflexive thematic analysis draws on two sources: (1) participant talk during interviews and (2) field observations of ASR use in clinical settings.</p><p><strong>Findings:</strong> The analysis shows how ASR, introduced as a mature, “trainable” AI, redistributed coordination work from secretaries to clinicians, who became responsible for safeguarding accuracy and accountability in health records. Clinicians performed invisible work correcting errors, managing breakdowns and making the system “livable.”</p><p><strong>Practical implications:</strong> The findings highlight the need for health-care organizations to recognize and resource the work required to sustain AI-supported documentation.</p><p><strong>Originality/value:</strong> The research paper demonstrates how ASR depends on clinicians’ everyday corrective actions and sensemaking to function. By examining the long-term persistence of the technology in practice, the study shows that AI-supported documentation is not a one-time implementation but an ongoing sociotechnical accomplishment shaped through clinicians’ interpretive and corrective labor.</p>