News
Here we are, less than a year after that Summit, and AI-based medical scribe tools have first astonished us, and then been implemented by many as a part of routine clinical workflow, particularly in private outpatient clinics. The potential for them to be implemented in public and private hospitals to reduce the burden of documentation, and potentially improve the accuracy and completeness of certain non-clinical types of documentation is enormous.
Is there any potential downside to the implementation of human or AI scribes? The production of succinct, high-quality medical documentation and communication between providers is something that all doctors should be competent to do. Does the quality of documentation by junior doctors benefit from “cutting their teeth” writing thousands of inpatient notes and discharge letters, refining this skill as they progress? Without this, will they be sufficiently able to oversee the work of an AI scribe when they are consultants?
An even bigger issue could be the loss of the cognitive processing that occurs whilst notetaking. There are many different styles of learning and processing information, and I have no expertise in cognitive science; but for me, I know that if I sit in an hour-long lecture and don’t write anything down, I will generally retain close to nothing that I can recall a week later. I suspect that if, as an intern, I had hangrily followed a ward round of 30 inpatients and hadn’t written anything in the notes, I would be similarly clueless an hour later. I wonder if this would also affect my ability to structure and consolidate my clinical learning in the longer term. In other words, perhaps there is something formative in the process of notetaking that could go missing with the use of AI scribes in the early postgraduate years.
This is not to suggest that we should not push ahead with every possible option to unburden junior doctors from the huge burden of unproductive, inefficient administration that faces them every day. We should – and AI will win the race against humans for many tasks. Let’s hope the health system has the agility to adopt it within three decades of every other industry. Perhaps the temptation of AI-enabled fax machines will get it across the line.