Elsevier Medical Education presents an interview with Sarah Deacon, Senior Lecturer in Nursing and Healthcare, Nottingham Trent University, on the evolving role of AI in nursing education through clinical practice.
Recent data suggests that nearly 88% of higher education students are already using generative AI. In your experience at Nottingham Trent University, how is this "bottom-up" adoption affecting the way nursing students approach their clinical skills and applied practice modules?
One of the things that we’re noticing recently is that students are arriving with ChatGPT-like habits already formed. So, people using ‘record and capture’ and using AI to summarize lecture notes and so on, is becoming more common and they’re coming a bit unstuck when it comes to doing the practical tasks in clinical sessions because you can’t ‘AI’ a physical practical task. For example, in our first year, we’ve got a multiple-choice question (MCQ) exam, which has also got short answer questions around clinical skills, and you can use AI to prep for that. Whereas, when doing a clinical objective structured clinical examination (OSCE), it is difficult to use AI in that, so it is not an issue within that, to a certain extent. But what we’ve seen is a lack of that on-the-spot knowledge and there is this risk that because generative AI bypasses those building blocks for clinical reasoning development, we’re seeing that maybe that skill isn’t being developed and expressed so much.
We really need to make sure that AI literacy is part of clinical skills and practical skills modules in that students are going out into clinical areas where AI will be woven into the digital tools that they’re using on the ward. But we need them to have those clinical reasoning skills to be able to spot where AI has made an error, for example, and to trust their nurse’s guts. That’s something that I think over-reliance on AI limits the development of – that kind of clinical reasoning and instinct.
"I also think it’s really important that, as lecturers, we are seen to model that curiosity and that interest about AI. I called it criticality, but it’s a genuine curiosity about how it could help but also what its limitations are and to not come across as fearful or as anti-AI because it is already such an inherent part in clinical practice and our students need to feel confident and capable that they are able to use it in the healthcare setting that they’re going into."
As a Senior Lecturer, how do you believe academic institutions should support faculty members to stay updated with AI tools that evolve on a weekly basis, ensuring they don’t fall behind their students?
I think it’s really important that institutions get on top of this and give dedicated time to it, instead of faculty just being told about a webinar that’s available in a lunch break now and again. We need to actually grip it as an issue and build it in. The most effective things I think that we’ve done within our department have been to have peer-led teaching sessions where people who are passionate about AI, and perhaps have more insight into it because of their interest, have been able to share their findings. For example, their findings on different tools that have come out that are really useful from a lecturer’s standpoint, but also on tools that students are now using that bring about issues around academic integrity. And actually, what is really important is developing staff members’ criticality around the processes as a whole. So, not just saying, “Okay, we need to look out for this particular program,” but actually, reading students work, for example.
I’m one of the Academic Integrity Leads and it’s becoming more and more difficult to spot the pieces of work that are completely produced by generative AI software. We need to be teaching lecturers to look for the principles rather than the nitty gritty, because you could teach the nitty gritty but it changes every day, doesn’t it? So, it’s about the overall criticality rather than the specific details about specific tools.
I also think it’s really important that, as lecturers, we are seen to model that curiosity and that interest about AI. I called it criticality, but it’s a genuine curiosity about how it could help but also what its limitations are and to not come across as fearful or as anti-AI because it is already such an inherent part in clinical practice and our students need to feel confident and capable that they are able to use it in the healthcare setting that they’re going into. But also, with that knowledge that they have a safe clinical background themselves and they’ve got that reasoning and that ability to question any AI outputs because they’re confident in their knowledge base as they’re viewing that output.
As generative AI becomes more common in nursing education, what are the inclusivity considerations educators should be aware of and what support or adjustments are needed to support equitable education for all nursing students?
I think the inclusivity aspects of using AI tools just mirror daily life, really. For example, when we think about neurodiversity, Gen AI can really level the playing field by reducing cognitive load and helping students who struggle with viewing a large task to chunk it down and make things easier. And I’m sure people are writing pedagogy about how it can level that playing field. However, when applying that in practical terms, if we have students with a neurodiversity, for example, who’ve been told that they can use the paid version of Grammarly, but students without a diagnosis are told that they can’t, it means that very different types of work is being produced. It produces inequality as the students without a diagnosis feel it’s not fair that other students are allowed to use Grammarly to produce their work. At the same time, the students using Grammarly may be accidentally sent through an academic integrity process because it’s determined that they’ve been using unacknowledged generative AI to produce their work.
Also, the majority (over 50%) of our nursing students are mature students. And I don’t want to stereotype but generally, there is a reduction in the level of digital skill for mature students as they enter university so they might already be on the back foot. So, in terms of accessing and using AI, they’re already having to make a much bigger step change in order to fluently use those AI tools.
We have a lot of international students and again, these large language models can be really useful for students where English is not their first language. But behind that comes a real risk of it masking a lack of cultural competence and when you’re teaching students to become nurses and to be qualified in a cultural or care setting, that level of knowledge is really, really important. So, if they’re relying on AI tools to produce their work, but it’s masking a lack of cultural competence, it can be a real worry.
The last thing is about digital poverty. Something I really noticed a couple of years ago when marking students’ work was the difference between students who were using the free version of ChatGPT versus the paid version. The students who were using the free version were getting picked up all the time and sent through academic integrity processes because of the age of the sources that they were using. It was a real tell because the free version couldn’t get behind those paywalls, so it was using significantly older citations and they were getting picked up. Therefore, instead of levelling the playing field, there is that dichotomy between rich and poor again. Digital poverty is extending through and is then going to affect the output for those students as well.
You operate as an expert witness for clinical breach of duty. From a legal and accountability perspective, what risks do you foresee if healthcare professionals rely too heavily on AI-derived information for clinical decision-making?
I’ve actually come across this in my role as an expert witness. It was reported that an AI-generated opinion had been relied on in somebody’s liability statement that they produced and it turned out that the references that they were using were hallucinated. So, it called into question their integrity and completely undermined their legal integrity in what they were submitting to the courts.
From an education point of view, we now have an inherent obligation to instill that criticality into our students. We need them to know that AI isn’t always right. We need them to feel confident enough to challenge the output and say, “Actually, is this correct?” and to know where to look up the sources that AI is citing so they’re aware that it may well be a hallucination and they’ve got the ability to question it and find out the truth. Ultimately, if they are able to do that, AI can be a really useful tool and they can feel confident that the output is safe and that it won’t have any implications for them in terms of accountability of their practice.
Given your research and publication work in palliative care, what do you consider the most significant ethical challenge of introducing AI into end-of-life care discussions or patient support?
It’s a really interesting question and I can never really imagine AI being part of end-of-life care. Obviously, tools are going to be used. For example, you might get it involved when clinicians are working out prognosis based on various risk factors and so on. But, in terms of actually delivering care within a palliative care setting, I’ve always said this – nursing is about being with somebody through something and you can’t do that other than being with them. AI can’t do that. There isn’t a way to have a conversation around end-of-life care that AI could take because so much of nursing is about what is not said, and AI tools rely on what is said. So, it will only use information that it is given. It can’t use information that it’s not given. And for me, nursing and creating your clinical judgment is about understanding what’s not said as well – what information is not there. All of those pieces go together to create that picture that then enables you to be with that patient through that process. Sitting with those uncomfortable feelings through an end-of-life conversation is so, so important and is something that, particularly within end-of-life settings, AI could never replace.
What role do you see for librarians in guiding both staff and students on the selection of AI tools that are not only effective but also compliant with healthcare data protection standards?
I think there’s an amazing role that librarians are going to have to take on in the future. To be honest, I need to have a chat with our librarians about this because I’m not sure whether they’ve started to do this yet, but I can see there being a need for them to start doing workshops with students around prompt generation, for example. Building that kind of literacy with AI to be able to get the most out of it and make the most of it in that critical way. I think librarians are going to be really essential with maintaining the veracity of output that students come up with, as well as researchers as they go on in their careers.
In terms of making sure that safe AI tools are promoted – that would be a really important role for librarians. Getting on top of what’s out there, helping students use AI in a positive way, and making sure that prompt literacy is taught alongside digital literacy, is going to be absolutely key, I think.
"So, I would say don’t be fearful of AI – question it, be curious, but also be excited because there is so much potential for assessments, for example, to be much more authentic and, actually, much more useful for quite a lot of students going out into the real world of clinical practice."
What is your message to nursing professors and librarians who feel distrust or fear that AI might erode the professional identity or the “human touch” of the nursing vocation?
I can see that people would have that fear. Generally, when I’ve come across people having that fear or that anxiety around it, it’s been about the validity of the students’ qualification, because if they have generated all their essays using Gen AI, what use is that piece of paper that they come out with at the end? Whereas actually, that is then on us as academics to reflect and say, “Okay, what do we now need our students to come out with?” And ultimately, it’s about synthesis of information, isn’t it? It’s about being able to assess, evaluate, and synthesize the information that they have all around them, whether that’s out of a journal or the patient in front of them, and to be able to make a judgment on the back of that.
We used to do that with essays. We used to ask the students to come up with an essay because it showed that they were able to evaluate information, critique it, synthesize it, and then produce it into an essay. We can’t do that now. Essays are dead and it’s up to us, as academics, to come up with different ways to allow our students to develop the skills and to show us that they’ve developed those skills.
So, I would say don’t be fearful of AI – question it, be curious, but also be excited because there is so much potential for assessments, for example, to be much more authentic and, actually, much more useful for quite a lot of students going out into the real world of clinical practice.
Looking at the future, what is the unique human skill – the one that AI cannot replicate – that you urge your students at NTU to strengthen most?
Essentially, the thing that AI will never be able to replace is a nurse’s gut. That feeling of ‘there’s just something not quite right’. That intuition when you look at a patient – we used to call it “non-specific ‘iffy-itus’” when I was newly qualified – and you just knew that something was off. Even though the numbers all looked fine, something had changed. I would say that all nurses have that ability to know when something’s not quite right, based on either years of experience or even by just knowing that patient for a couple of hours. And that’s something that AI will never be able to spot.
As I said, nursing is about being with somebody through something. My background is in the operating theatre so I used to be a cardiothoracic surgical care practitioner and for me, acting as a patient advocate within theatre was just as essential as doing the task that was in front of me. Also, acting as an advocate through whatever your patient is going through, again, relies on the whole picture and information that might not necessarily be particularly obvious. And making those clinical judgments is something that artificial intelligence will never be able to do because it relies on information that might not be present as well as the information that is.
Leave a Reply