Artificial intelligence tools designed to assist physicians with note-taking may overlook vital information during patient consultations, according to a study from the University of Edinburgh. These AI scribes, while capable of capturing clinical details, can miss non-verbal cues such as facial expressions, gestures, and the patient's emotional state.
The research suggests that AI summaries tend to prioritize clinical data, potentially at the expense of patients' personal stories and their experiences of illness. Furthermore, patients might be hesitant to disclose sensitive information, including substance abuse, domestic abuse, or mental health struggles, when aware that their consultations are being recorded and processed by AI.
Researchers also highlighted the importance of manual note-taking for clinicians, noting its role in their reasoning and reflection processes. The use of AI scribes could lead to "cognitive offloading," where the tools assume some of the mental effort traditionally undertaken by clinicians. While this can free up physicians to engage in more meaningful conversations or focus on complex tasks, it may also diminish memory recall and skill development.
Instances have been noted where clinicians did not recognize their own notes or recall patients during subsequent visits, indicating a potential long-term impact on their clinical acumen. The report concludes that while AI scribes can aid with administrative tasks, further research is necessary to understand their efficacy within specific healthcare settings and how their design can best support both patients and clinicians.
Despite these concerns, AI transcription tools have shown potential benefits in reducing administrative burdens. At one clinic in Dudley, a transcription tool named Heidi was credited with reducing a backlog of patient letters from six months to 14 days, significantly cutting down on doctors' workloads.