Nurses as Reference Points in the Age of Artificial Intelligence
Author: Dr Doina Carmen Mazilu1,2
1. JBI Romanian Centre for Nursing Research
2. OAMGMAMR filiala Municipiului București
A phenomenon I am noticing more and more is that the patients arrive at the hospital with information already generated by artificial intelligence (AI). However, they still require a human conversation in order to understand what that information truly means for their personal situation.
What surprises me is how quickly this has become a common situation in healthcare: a patient arrives with a smartphone and questions already formulated with the help of AI. Their presentation may seem clear and convincing, but in real-world healthcare, that does not always mean it is accurate, complete, or safe to treat at face value for that individual.
For nurses, this is not merely a technological issue. It is also a matter of communication, trust, and, sometimes, safety. What I find particularly important is that nurses have always worked as translators at the intersection of medical knowledge and everyday life. To me, this is one of the most valuable dimensions of nursing practice: explaining recommendations, helping patients understand what to do at home, noticing any confusion, and translating clinical language into something practical and realistic. This important function is one reason why I believe nurses remain essential in the age of AI.
What AI still cannot do is be physically present at the patient's bedside. A nurse can notice when a patient is breathing more laboriously, or when they become more confused, withdrawn, or unstable. At that moment, a nurse can detect the subtle changes that may signal deterioration, and they can respond promptly when it is necessary. This type of attention is not an additional element of care, but a primary role of the healthcare professional. It may be key to the safety of the patient and cannot be replaced by a response generated on a screen.
AI may change the tools that support this work, but it cannot take it over. Nor should the advent of AI undermine the importance of this interpersonal work from nurses.
From my perspective, AI can be genuinely useful for communication in healthcare. I understand why these tools appeal to busy professionals: they can simplify technical language, help draft patient education materials, and summarise long documents. Nurses are often expected to communicate important information quickly, clearly, and safely, so tools that promise to save time naturally attract attention. However, speed is not the same as safety.
In healthcare, safe communication does not mean only the fast transmission of information. Beyond that, it means recognising what the respective patient understands, what they need, what worries them, and what they may fail to notice or fully understand.
An AI tool does not know the patient as a whole. It cannot read a facial expression, hear hesitation, or understand the context behind a question. It may also miss local realities, such as available services or health literacy. More importantly, it can generate information that sounds professional even when it is not reliable, up to date, or evidence-based, despite the growing attention to ethics and governance in the use of AI in healthcare.
This is the reason why I do not view AI as fit to replace nurses in the communication of evidence. Instead, I see it as a tool that can support nurses, and only when used critically, responsibly, and with a clear understanding of its limitations, in line with appeals for the responsible use of AI in evidence synthesis .
Rather than considering AI as an endpoint, a better way to view AI is as a starting point. For instance, an AI tool may assist in generating a plain-language explanation about wound care, diabetes prevention, or treatment adherence. However, before that explanation reaches the patient, it should be carefully reviewed by a nurse or another qualified healthcare professional. Is it accurate? Does it fit the current recommendations, including the principles regarding the use of AI in the development of guidelines? Is it appropriate for this particular individual? Is anything important missing? Could any phrase be misunderstood?
These are not merely technical questions. They are part of professional care.
Evidence-based nursing does not simply mean finding research; beyond that, it involves the judicious use of evidence in conjunction with clinical expertise and the values and preferences of the patient. As such, AI does not eliminate the need for professional judgement: it raises its importance.
When individuals can access large amounts of health information within seconds, they need assistance to decide what is trustworthy and what may be misleading. For this reason, I believe that a nurse’s response to a patient who has used AI should not be one of rejection. In most cases, that person is not attempting to challenge the care provided. Rather, they are trying to understand, prepare, or regain a sense of control. If I were part of that conversation, I would like it to begin with curiosity: ‘What did you find?’ or ‘Let’s look at it together.’ In my opinion, this approach preserves the person’s dignity and paves the way for correction without causing embarrassment.
This approach matters because disinformation is not merely a matter of false facts; it is also a matter of trust. AI-generated information can sound collected, fluent, and convincing. To a patient, this element of composure may feel reassuring. Yet to a healthcare professional, it should be a signal that action is needed to verify the source, context, and evidence underlying the information.
Thus, the nurse’s role becomes one of guidance. This does not mean they should stand between the patient and information, but rather help the patient make sense of that information. Instead of rejecting technology, the nurse is working to ensure that technology supports safe and humane care.
What Nurses Need Now
First, nurses need training in AI literacy. This does not mean becoming technical experts, but understanding the main strengths and limitations of these tools: AI can help with language and structure, but it can also generate errors, omit important details, or reinforce existing biases.
Second, healthcare organisations need clear rules on how AI-supported content may be used. In particular, patient educational materials, discharge instructions, and clinical summaries should not be copied directly from an AI tool and used without review. A simple process should be established to verify their accuracy, underlying evidence, readability, and local relevance.
Third, the nurses should be involved in the design, testing, and implementation of AI tools within care settings. This is important, as nurses understand the practical reality of patient communication. They know where confusion may arise, where instructions fail, and where even a technically correct answer may be insufficient. Without their contribution, AI tools may seem impressive but fail to match up to the realities of care delivery.
The same principle of co-design also applies to patients and communities. AI in healthcare should not be designed merely for the people, but together with the people. A tool that works well for one group may fail for another, as language, literacy, digital access, disability, culture, and trust all influence how health information is received and used.
Lessons Learned
From my perspective, what appears to work is using AI as a support tool for clearer communication, rather than as a substitute for professional responsibility. What does not work is assuming that a well-articulated response is necessarily trustworthy.
The lesson I always return to is that the human dimension of care becomes even more important when information is easier to generate but more difficult to evaluate. If we wish to use AI tools effectively in the future, we need more training, clearer governance, and a stronger involvement of nurses, patients, and the community in shaping how AI is introduced into healthcare.
At the patient’s bedside, evidence represents more than a mere recommendation from a document. It becomes a conversation, and a decision about what a person understands, accepts, and is capable of doing. This is where the care provided by nurses remains essential.
AI can help bring evidence closer to nurses. However, it is nurses who bring evidence closer to patients. This distinction matters. AI can improve speed, access, and clarity, but nurses provide professional judgement, accountability, and context. In evidence-based nursing, these qualities are essential to safe care. For nurses, the challenge is not simply to use AI, but to guide its use carefully.
Key Messages
AI can support nurses in communicating health-related evidence, but should never replace professional judgement
Nurses are essential human guides who help patients to understand, ask questions, and use health information safely
AI tools in healthcare should be verified, governed, and developed with the contribution of nurses, patients, communities
References
Cochrane. (2025, February 28). New AI Methods Group to spearhead adoption across four leading evidence synthesis organizations. https://www.cochrane.org/about-us/news/new-ai-methods-group-spearhead-adoption-across-four-leadingevidence-synthesis-organizations
Guidelines International Network. (2025). Principles for use of artificial intelligence in the health guideline enterprise. Journal of Clinical Epidemiology [advance online publication]. https://pubmed.ncbi.nlm.nih.gov/39869912/
World Health Organization. (2024). Ethics and governance of artificial intelligence for health: Guidance on large multimodal models. https://www.who.int/publications/i/item/9789240084759
To link to this article - DOI: https://doi.org/10.70253/ZWRI6313
Disclaimer
The views expressed in this World EBHC Day Blog, as well as any errors or omissions, are the sole responsibility of the author and do not represent the views of the World EBHC Day Steering Committee, Official Partners or Sponsors; nor does it imply endorsement by the aforementioned parties.
