Hasan Erbay
Health Sciences Quarterly - 2026;6(2):235-238
Something has shifted in medicine, and we are still trying to understand its implications. Artificial intelligence can now interpret a retinal scan with greater accuracy than most ophthalmologists. Algorithms flag sepsis before the nurse at the bedside does. Large language models draft discharge summaries, summarise referral letters, and answer patient questions at two in the morning. The efficiency gains are real. And yet, none of this touches what a patient actually needs when they are sitting across from a physician, frightened and uncertain. That gap is not a sentimental observation. It is a structural problem. As AI takes on more of the cognitive labour of medicine (pattern recognition, risk stratification, protocol optimisation), the capacities it cannot replicate become more, not less, important. Listening. Interpreting. Being genuinely present with a person in the middle of a difficult story. These are not what AI does. They are what physicians do. Or what physicians should do when the system gives them time and space to do it. Narrative medicine, as Rita Charon has argued for over two decades, is clinical practice anchored in narrative competence: the ability to recognise, absorb, interpret, and act on the stories patients bring to the encounter [1]. The framework is sometimes misread as a plea for softer medicine. It is not. It is a claim about what clinical knowledge actually requires. A diagnosis that ignores the patient's story is not a complete diagnosis. The distinction Charon draws (between the diagnostic gaze, which seeks what is measurable, and narrative attention, which seeks what is meaningful) is not a distinction between science and sentiment. It is a distinction between partial knowledge and whole knowledge. One way to see this clearly is through the old separation between disease and illness. Disease is what the scan shows. Illness is what the patient lives with. An algorithm can map the former with extraordinary precision. It cannot touch the latter: the fear of what the diagnosis means for one's children, the loss of a sense of self, what Pellegrino described as 'wounded humanity' (as cited in Johna & Rahman) [2]. Consider a patient who has been told, correctly, that her tumour is resectable. The AI got the biology right. But she is sitting in the consultation room thinking about her daughter's upcoming wedding and wondering whether she will be there. That thought is not in the data. It is in the story. And it matters - clinically, ethically, humanly. Narrative ethics presses this point further [3]. Whose voice actually shapes the treatment plan? When an algorithm offers a recommendation based on population-level data, someone still has to ask: Does this fit this person, with this history, in this moment of their life? That question cannot be answered by processing more data. It requires the kind of moral attention that narrative ethics cultivates: an orientation toward the patient not as a case but as a protagonist. Humane clinicians, one might argue, have always listened this way, always brought moral attention to the encounter, always read the patient as well as the chart. This is not entirely wrong. But that attentiveness has never been more threatened than it is now, and what is threatened needs defending. Here is a finding that should give us pause. Recent studies show that patients, in some contexts, rate AI chatbot responses as more empathetic than those of their physicians [4]. While algorithms can mimic the surface patterns of narrative, the empathy at the heart of medicine requires more than a calculated response; it demands a genuine encounter between the clinician and all those who suffer. This does not mean that machines have learned to care. It means that physicians (worn down by administrative burdens, time pressure, and the relentless demands of digital documentation) have been systematically deprived of the conditions that empathy requires. That is not a reason to hand empathy over to the algorithm. It is a reason to ask what we are doing to our clinicians, and to our patients, and to change it. Toward a new emphasis in medical education As a physician who has spent years teaching ethics and the history of medicine, I have noticed something. Students arrive with impressive knowledge and no small anxiety about the things they do not yet know. What they are less prepared for - and what the curriculum does little to address - is the disorientation of sitting with a patient who is not asking for information but for understanding. That disorientation is not a personal failing. It is a training gap. And it will only widen as AI takes over more of the information-management tasks that have historically structured the clinical encounter. The skills that AI handles best (recall, pattern recognition, protocol application) are precisely the skills that medical education has most vigorously trained. That was probably never the right emphasis. In the AI era, it is becoming harder to defend. What needs to come forward instead are the competencies that resist automation: empathic communication, ethical judgment, cultural attunement, reflective practice, and narrative understanding. A systematic review of narrative medicine programmes finds that such training reliably builds empathy, communication, and professional resilience, qualities no algorithm has yet managed to approximate [5]. These are not electives. They are the curriculum. Pedagogical models variously called narrative-based learning or story-based learning offer promising structures for this work. Through engagement with patient narratives, literary texts, reflective writing, and case-based storytelling, students develop the interpretive and empathic capacities that clinical encounters demand. A comprehensive scoping review of narrative medicine programmes concludes that such approaches are both feasible and effective, not as supplements to medical education, but as genuine contributions to clinical competence [6]. The question is not whether to include them. The question is why they are still treated as optional. My own published work in this area has tried to show that narrative medicine is not a theory one admires from a distance but a pedagogy one can practise, teach, and assess [7]. That work is ongoing. What I am more confident of now than when I began is that the integration of narrative competence into medical education is not a luxury for institutions with time and resources to spare. It is an ethical obligation, and one that becomes more urgent, not less, as AI reshapes the clinical landscape. Looking ahead The physician of the near future will work in an environment dense with algorithmic outputs, predictive scores, and AI-generated documentation. Some of this will be genuinely helpful. Some of it will be wrong. All of it will require interpretation: not just technical reading, but the kind of contextual, values-sensitive judgment that no system can yet perform on behalf of a human clinician. The physician who can do this well will need, as I have argued elsewhere, a kind of bilingualism: fluency in data and fluency in story. Reading a genomic profile and hearing a patient's fear are different skills. They are not in competition. But if we do not deliberately cultivate the second, the first will crowd it out. It is tempting to conclude that the medical humanities face an uphill struggle in this environment, that narrative medicine and narrative ethics will always be the poor relation of the hard sciences in a world that rewards efficiency and measurable outcomes. That conclusion seems premature. Recent scholarship makes the unexpected observation that large language models are themselves deeply narrative in nature, trained on vast stores of human story, generating outputs that mimic the structure and texture of human reasoning [8]. Seen this way, the medical humanities are not standing outside the AI era looking in. They are, in some sense, inside it, and better placed than any other discipline to evaluate what is being gained and what is being lost. None of this is settled. The questions narrative medicine and narrative ethics raise, about whose voice is heard, about what counts as care, about what we owe to patients as persons rather than as cases, will not be resolved by better algorithms. They will be resolved, if at all, by physicians, educators, and institutions that take them seriously. That is the work. It is harder than it sounds, and more important than it may appear. I hope this editorial contributes, in a small way, to keeping it on the agenda. Conflicts of interest The editor declares no conflicts of interest.
