Elsevier Medical Education presents an interview with Dr Ximena Alvira, Clinical & Research Manager at Elsevier Health, exploring how AI is reshaping health education and clinical practice, and why developing critical thinking in future clinicians is more urgent than ever.
How can AI help build the bridge between education and clinical practice, and better prepare future clinicians?
Building virtual AI scenarios that reflect real patients
I want to highlight two aspects here. One of them would be the creation of virtual scenarios, either through AI virtual patients or just the scenario setting to emulate the challenging situations that students are going to face as clinicians. As students, we learn from the perfect scenario and we prepare for tests that are the perfect scenario but rarely do these scenarios incorporate the real life into the patient. By real life I mean the patient having the various personal, emotional, financial, biological, psychological, geographical, and ethical determinants that really impact on how we, as patients, get better, get healthier, and recover from a specific condition. So, I think creating these virtual AI scenarios is going to be really important and interesting because it brings evidence from such an enormous corpus of data from around the world that we can use to curate the truly diverse settings of real-world clinical practice.
Critical thinking: how AI can sharpen or erode it
The second aspect where I think AI can help build the bridge between education and clinical practice is through critical thinking. However, it has the potential to help or impair critical thinking. AI can help guide us through the exact reasoning process until we reach a specific clinical decision and highlights what the pros and cons of making those decisions are. But then there is the risk that we start to over rely on AI and forget about our critical thinking. So, hopefully it’s the first option.
"Exercising and fostering critical thinking needs to become a renewed area of brain development among students because, although this skill has always been there, it needs to be prioritized further to avoid over relying on AI."
Dr Ximena Alvira, Clinical & Research Manager, Elsevier | Elsevier Faculty Hub, June 2026
What core competencies related to the use of AI should students have when entering the clinical environment?
Fostering critical thinking against over-reliance on AI
As I mentioned before, critical thinking can be enhanced by AI because it guides us through reasoning processes before we apply it to a patient, but this can also be used in the wrong way. We could become overly reliant on AI and forget that we are ultimately responsible and accountable for everything we do for our patients and in the healthcare setting that we’re working in. Fostering critical thinking is of utmost importance and is increasingly so because of this ability to use AI with such confidence and trustworthiness that we rarely criticize it or consider that it might be just plain wrong. Exercising and fostering critical thinking needs to become a renewed area of brain development among students because, although this skill has always been there, it needs to be prioritized further to avoid over relying on AI.
AI literacy as a non-negotiable core competency
Secondly, I believe that AI literacy is a core competency. Understanding what the limitations are, what the challenges are, what different AI models are being used in healthcare, which ones have proven to be safer for our patients, and so on. Having a profound literacy on AI is non-negotiable anymore and it should be incorporated very early into our careers because this is what students are going to use once they get into clinical practice.
So, I would say that these are the two main core competencies, but of course, there are many more such as legal aspects or privacy issues that can be grouped under these two competencies.
From a broader perspective, how is AI currently being implemented and used in hospitals by residents and clinicians?
Two decades of AI in clinical decision support
AI has been used in hospitals for many years now – for more than two decades. So, it’s nothing new. It’s being used for clinical decision support by retrieving information in a fast, efficient, and safe way. It’s also being used to support diagnosis across a range of specialties – in Radiology, for example, and in the early detection of things like diabetic retinopathy and strokes. These are prediction models that use machine learning to ultimately improve patient care.
The next step: truly personalized patient care
I would say that the latest development has been the integration of real-world patient data with already defined and validated AI models, built with evidence-based research, and combining these two datasets to propose a care approach that is extremely personalized for a specific patient, a specific context, and a specific scenario. That will be the ultimate delivery of personalized care, which I’m very much looking forward to seeing in practice.
So yes, AI is being widely used. We know large language models are also being used – albeit not very safely – for retrieval of information as well as diagnostic support. AI tools are becoming great allies for clinicians and residents alike to support them in their daily work.
How can medical education ensure continuity between how AI is taught during training and how it is actually used in clinical practice?
Aligning training with real clinical workflows
I can think of several ways for how medical education can ensure continuity here. I think aligning with the real clinical workflows that any healthcare providers are going to encounter during hospital training or development. This is one of the most important ones because I still see this gap of what we learned at medical school and then what we’re forced to learn in a very fast and sometimes painful way when we are practicing as clinician. So, ensuring that continuity because we will be using similar or the same tools in clinical practice and this ensures that we won’t have that gap.
Teaching students to evaluate AI tools critically
Also, we should be fostering critical thinking, as I previously mentioned. Not critical thinking about the AI tools themselves, because they will continue to evolve so fast that we won’t be able to keep up but rather, about the skills students will need as clinicians to be able to recognize when an AI-generated output is worth applying or abandoning. Students need to be taught how to evaluate these AI tools, almost like a Health Technology Assessment approach. Because this is what students will face when they move into clinical practice and whether they make the right choice or the wrong choice will depend on how good they are at evaluating its risks and its benefits. For me, critical thinking is about ensuring continuity and early exposure within a student’s career to the AI tools that clinicians will be using once at a professional level.
"Students need to know not only what AI tools are being used [in clinical practice] and how, but also, what their challenges are and what the ethics and accountability factors behind them are."
Dr Ximena Alvira, Clinical & Research Manager, Elsevier | Elsevier Faculty Hub, June 2026
In your opinion, what are the main challenges or gaps that still exist in integrating AI effectively across the continuum from education to clinical care?
The gap between training and clinical reality
I would say there are two very important challenges and gaps. One is the training and practice gap because most medical students are being taught about AI, but they’re not told how AI is being implemented in hospital or clinical practice. So, there’s this lag or gap where students think that the use of AI tools is about answering tests or writing a thesis but then, when they go into clinical practice, these AI tools are being used for diagnostic support, for image recognition, for clinical decision support. Students need to know not only what AI tools are being used and how, but also, what their challenges are and what the ethics and accountability factors behind them are.
AI adoption is outpacing policy and accountability
This moves me onto the second problem, which is that AI tools are being adopted faster than the speed at which policies are being created. There needs to be a high degree of accountability before we use AI tools, but once we know which tools should be used, we can better teach how they should be used. It’s not about teaching the technology behind the tools, but rather, teaching what the constraints are, what the good uses are when deployed, when they are poorly deployed, what the challenges are, and what the consequences can be. For a concrete example, students learn from books that we can trust, but when we’re faced with these AI tools in clinical practice, what are we going to trust and how are we going to evaluate them? So, for me, an existing gap is how we can trust what we read.
Leave a Reply