Dr Laura Green, Senior Lecturer and Programme Director in Nursing at the University of Manchester, asks whether AI in nursing education is a helpful tool or an existential challenge and argues that regulatory bodies must be part of finding the answer.
The conversations I am having about artificial intelligence in education are no longer about saving time on PowerPoint slides or summarising long documents. They are about something unsettling: whether the cognitive foundations on which we have built professional nursing education are fit for a world in which AI can do the thinking for us.
Recently, Anthropic withdrew their latest LLM model after the US government released a directive preventing its release. Whilst that is not the subject of this post, it is relevant because it points a finger directly at this brave new world we are navigating.
I should be transparent. I did not use AI to write this post, though I have used it to churn ideas, test arguments and cut my word count. I sit with a combination of guilt and excitement about what it offers, and I am writing this post to try and articulate why I think this ambivalence is acceptable. We will not find solutions to wicked problems if we polarise and position before we engage in the debates.
Across my work at the University of Manchester, there are many examples of use of AI by academics. I have been building and testing AI applications that go beyond summarising lecture notes. The InterACT-AI project brings together nursing, medicine, pharmacy and speech and language therapy students in interprofessional simulation scenarios. Academics can now vibe-code web apps that reflect authentic, high fidelity Electronic Patient Records (EPR) to provide students with realistic clinical learning on campus.
Educationally, this is a phenomenal development, that nursing (and all other regulated professional programmes) stand to gain significant benefits from. In the classroom, students have traditionally practised clinical reasoning in a kind of information vacuum: paper charts and verbal handovers that bear little resemblance to the rich, messy, multi-layered data environments of real wards. The simulated EPR help students navigate the same complexity they will face in practice, and educators can observe, adjust and debrief in real time.
In nursing, there are clear criteria regarding what constitutes simulated practice learning and higher education institutions are under pressure to build and evaluate scaleable approaches to simulation, minimising the use of expensive software or technology. As educators we now have the ability to draw on the capabilities offered by AI and create resources that are tailored to our students and our programmes.
Outsourcing vs offloading
There is a difference between outsourcing cognition and offloading it. Outsourcing is strategic and deliberate. A nurse using a medicines calculator to check a drug dose calculation they have already reasoned through mental is not avoiding thinking but verifying it. The tool serves the nurse’s cognition.
Offloading is what happens when the tool substitutes for thinking that was never done in the first place. When a student asks an AI to generate their reflection, interpret their patient data or summarise the evidence for a clinical decision and accepts the output without engaging critically they are not extending their mind.
This distinction maps onto a growing body of work on what researchers have begun calling ‘metacognitive laziness’, or the tendency, when powerful tools are available, to disengage from effortful thinking because the tool makes it unnecessary (Bastian et al., 2024; Gerlich, 2025). The risks include academic underperformance and patient safety and they are profound. The literature on cognitive load theory, extended mind theory and distributed cognition all show that human beings are natural cognitive offloaders (Clark & Chalmers, 1998; Sweller, 1988).
In most domains, this is just seen as human progress. The Nursing and Midwifery Council (NMC) requires that registrants must be able to demonstrate knowledge, skills and professional values in their own right. They cannot delegate their reasoning to an algorithm. They will be held legally and professionally accountable for clinical decisions. The NMC registers practitioners who can demonstrate, independently and under pressure, that they know what they are doing and why. This is the paradox that concerns me and occupies many waking hours. We are training students to work with tools that may erode the very competencies that make them safe practitioners.
We need to ask whether our professional programmes are preparing students to understand and navigate these changes with integrity. We need our regulatory bodies to step into this space with authority and collaborative intent. The NMC sets the standards by which nurses are registered. Those standards were developed before generative AI was a reality in education. They need revisiting to articulate with far greater precision what ‘demonstrating competence’ means in an AI-mediated world.
We now need collaborative policy development where we bring together academics, regulatory bodies, professional organisations, patient advocates and students. The NMC, the General Medical Council and the Health and Care Professions Council are each grappling with versions of this challenge across their registrant communities. There is a strong case for cross-regulatory dialogue to help build a shared framework that distinguishes between AI as a legitimate learning scaffold and AI as a competence proxy. This would be welcomed by educators, who are in serious need of clear and defensible ground on which to stand.
Without that infrastructure, individual academics are left making high-stakes, potentially inconsistent, decisions about academic integrity, while the deeper structural question of what it means to be a competent professional in an AI-enabled healthcare environment remains unanswered.
I want my students to engage with AI as a thinking tool that prompts, challenges and supports. We need to build relational and reasoning skills that remain human. We also need to encourage and retain cognitive discomfort and AI risks making this process too comfortable. In case you had not guessed, I am decidedly not anti-AI. Far from it. I believe these technologies offer transformative possibilities for healthcare education – possibilities we are only beginning to understand. But I am deeply wary of the tendency in higher education to adopt AI enthusiastically at the surface (generating lesson plans, summarising readings, drafting feedback) while avoiding the harder, structural questions about what our programmes are actually for.
For professionally regulated programmes, those cannot be answered by educators alone. We need regulatory bodies at the table as co-architects of the policy environment in which AI-era professional education takes place. The question is not ‘How do we use AI in our teaching?‘ The question is ‘What do we believe nursing education is fundamentally trying to do and is AI helping us achieve that: or is it doing something else entirely?’
I would love to hear how others are navigating this tension. Are your students outsourcing their thinking or extending it? And do you feel you have the regulatory and institutional support to respond?
Dr Laura Green is Senior Lecturer and Programme Director in the Division of Nursing, Midwifery & Social Work at the University of Manchester. She has an interest in digital education, AI in nursing and simulation-based learning, and is co-investigator on the InterACT-AI interprofessional simulation project.
References
Bastian, L., Karatas, H., & Stöckli, S. (2024). AI tools and metacognitive laziness: Implications for higher education. Computers & Education: Artificial Intelligence, 7, 100254.
Clark, A., & Chalmers, D. (1998). The extended mind. Analysis, 58(1), 7–19.
Gerlich, M. (2025). AI tools in society: Impacts on cognitive offloading and the future of critical thinking. Societies, 15(1), 6. https://doi.org/10.3390/soc15010006
Nursing and Midwifery Council. (2023). Future nurse: Standards of proficiency for registered nurses. NMC.
Sweller, J. (1988). Cognitive load during problem solving: Effects on learning. Cognitive Science, 12(2), 257–285.
