The Human in the Machine: Reflections on AI and Professionalism

The recent College bulletin discussed the use of AI in histopathology covering the responsible use of AI tools in our specialties as well as the mix of fear, optimism and caution as AI use increases. There is no doubt that many aspects of our roles can be augmented or simplified by AI and our frustrations of repetitive processes eased.

With this in mind I asked Google AI tool to write a version of this blog using the following prompt: “Please write an 800 word blog in the style of Dr Christopher Tiplady discussing the use of AI tools in creative writing in relation to professionalism in doctors”. Any of you can enter that exact same prompt and will get a version.

I have that blog right now on my second screen and I will now add one paragraph from it to this blog. Can you tell which one?

As someone who supervises a Medical Education qualification I see other aspects of AI, like the use of it as both a research and writing tool. Much like we are expected to do in medical practice, we ask for reflective, critical writing as part of our course. We want people to experience teaching and learning, to reflect on the evidence behind it and to develop their own teaching and supervisory practice in response to what they see and learn. Part of academic writing is also that you can do it in a way that others understand and follow. It’s very easy to ask an AI tool to do all this for you, probably quite tempting too. We do spot the use of AI here and there, sometimes a bit too much, but that could just be our suspicious minds. If one too many em dashes appear we raise that metaphorical eyebrow.

Can we really tell when an AI tool has produced a piece of creative writing and does it matter? Considering what the GMC is guiding us to do, is it ethically and educationally wrong to use AI in your own reflective practice? Are you learning from this? Are you learning how to communicate with others properly or not? One of my Masters students is conducting a research project on the use of AI in doctor’s revalidation portfolios, I look forward to her results in the next year.

If a machine writes my reflections, it isn't my uncertainty I'm managing. It's a sanitized, sophisticated-sounding hallucination. A "pleasant" version of the truth that lacks the raw, honest insight gained from actually sitting with a patient and watching their face as you deliver earth-shattering news.

Medicine is a unique mix of social and natural science. We look after people and always want to apply the best evidence to their care, the issue is defining what “best” means. That is the social bit, that is where reflecting helps you manage uncertainty, complexity, people, systems and the good or bad days. It is where communication skills are learned. Leaving all that to a large language model will not help you personally and failing to share that reflection will mean no one else learns either.

When we interact as pathologists we need precision in our reports but nuance in interpretation and explanation of information. For example, watch Reddit for a while, and follow conversations from America where patients are highly alarmed by automated blood film reports. These could no doubt be regarded as much more accurate than my own, but phrases like “Burr Cells 1+” or “Schistocytes – Slight” just cause anxiety.  Defensive explanatory sentences are also added to any result a whisker outside of normal. Can you imagine receiving those reports, which are all strictly correct, but lack any humanity to determine what is important or not?

“The reflective practitioner - guidance for doctors and medical students” is something to read if you want to learn more about reflection with its very clear links to the GMC revalidation processes. We should all learn how to transfer information well and it is only through our own honest reflections that we will do this properly, succinctly and in a way that communicates complexity, nuance and relevance.

As for which bits of this blog are out of the machine rather than me – it is the paragraph beginning with “If” and the blog title. The rest of the AI blog was just so cheesy, maybe you will type the prompt into Google and see another version for yourself. I just asked for another one myself and found this sentence:

“If I use an AI to write this blog, I haven't actually reflected. I haven't cycled through the drizzle and let the ideas settle.”

It reminded me, I always try and get something in my blogs about cycling. Thank you Google AI.