Key points are not available for this paper at this time.
remains a concern, as the training data and decision-making processes underlying model outputs are often not visible to clinicians, raising the possibility of unintended biases.Such biases can limit the generalisability of AI scribing algorithms and perpetuate health disparities in underserved or minority populations 20.Finally, AI scribes transcribe sensitive patient information, which raise concerns for data privacy and security such as data breaches, unauthorised access, and misuse.These concerns emphasise the need for data protection, patient consent and transparency in the mechanisms underlying AI models 19,21,22.Artificial intelligence-based transcription tools mark a major advancement in modern ophthalmology, with uses in clinical care and operating room procedures.AI scribing increases ophthalmology workflow efficiency, quality of patient interactions, and patient comprehension in a high patient volume specialty.However, their use entails many limitations that necessitate strict physician oversight and robust privacy protections.This rapidly growing technology will transform ophthalmic medical documentation in the future, highlighting the need for increased discussion and further research into the safe and effective integration of AI scribing tools within ophthalmology.
Wang et al. (Wed,) studied this question.