The Irreducible Encounter: Human Skills, Artificial Intelligence, and the Future of Ophthalmic Care Authors: Tariq M. Aslam Published: 2026 Each day, in eye clinics worldwide, patients receive diagnoses that alter how they see their futures. To the patient, these are more than clinical events and may involve confrontations with loss of independence, identity, and the visual world as they know it. How clinicians handle these moments, by listening, explaining, acknowledging fear, and offering honest reassurance, can profoundly shape what follows in terms of psychological adjustment, treatment engagement, and ultimately patient outcomes [1]. The emotional weight of these encounters does not diminish after diagnosis. Anxiety and depression are highly prevalent among patients with chronic eye disease. A recent meta-analysis of patients with irreversible vision loss reported high rates of both conditions and concluded that there is a critical need for more holistic care [2]. Despite this, discussion and research into emotional and interpersonal dimensions of ophthalmic care remain conspicuously under-resourced, with the field’s imagination preferentially and perhaps understandably captured by pharmacological and technological advances such as in artificial intelligence (AI). Of course, the achievements of AI are highly impressive, across all fields of diagnosis, screening, grading, and prognostication [3, 4]. Indeed, the question is no longer about whether AI can replicate clinical judgment, but whether anything at all in the human encounter should be irreducible and protected from algorithmic decomposition. This is less a question about what AI could replace, and more about what it should not. One response to that question might be considered alongside our recent publication in this journal proposing Post-Visual Diagnosis Distress (PVDD) as a descriptive framework for early, diagnosis-linked emotional responses [5]. PVDD is not a psychiatric diagnosis but a construct drawing on evidence from age-related macular degeneration, glaucoma, and other chronic eye diseases, and informed by ophthalmology, psychiatry, counseling, patient advocacy, and public health. It legitimizes patient experience, opens space for empathetic dialogue, and enables timely recognition when distress crosses into something requiring more structured support. Its introduction reflects a broader emphasis on the emotional dimension of ophthalmic care deserving rigor and investment equal to that directed toward imaging and algorithmic innovation. PVDD illustrates that in clinical practice some things should resist automation not because machines lack the power, but because human encounters are more suited. Recognizing and managing the complexities of a patient’s distress is one such process, that is best suited to calibrated, relational human understanding. This key strength of human capability lies in the management of the psychological factors that shape clinical encounters. Procedural anxiety, for example, can be a significant issue in intravitreal therapy. Ziemssen and colleagues recently provided a structured, evidence-based framework for this context, emphasizing listening skills, avoiding catastrophic language, and adapting explanations to individual needs and anxieties [6]. A further inherently human skill required in the age of AI is in translating AI probabilistic outputs into language that is meaningful, honest, and responsive to individual values [7] When an algorithm estimates an 80% probability of an active neovascular membrane, how should a clinician combine this with their own opinion and communicate it to an anxious patient who experienced prior adverse events and wishes treatment only if absolutely necessary? The clinician serves as an essential interpreter: transforming statistical abstraction into language that communicates what the patient actually needs to understand in order to decide. The same interpretive responsibility applies in the converse situation, when an AI output suggests an 80% probability of inactive disease [8]. The psychology of this human/AI interaction deserves further study. McNamara and colleagues have shown that AI output characteristics, including the form of the core prediction and the presence and type of explainability, vary substantially across clinically deployed devices, placing considerable interpretive burden on the clinician [7], consistent with evidence that less experienced clinicians are particularly vulnerable to over-reliance [9]. These are not only technical problems requiring technical solutions. They are human problems requiring training and critical thinking about when to trust the machine and when to trust oneself. The rapid integration of AI into ophthalmology is something to embrace, but with a transformation that makes these human skills more prominent, not less. Much current AI research targets clinician-level decision-making such as detecting disease, grading severity, predicting progression [4]. These are significant achievements, but they address only a subset of clinical work. Administrative and documentation burdens that consume a substantial proportion of the clinical day remain comparatively under-addressed by AI innovation. The danger is that AI development focused on diagnostic replication rather than administrative liberation could leave clinicians with less time for human connection, not more. When documentation consumes the clinical encounter, clinicians experience what authors Dean and colleagues have termed ‘moral injury’, as a result of the accumulated distress of being systematically prevented from providing the care they entered medicine to deliver [10]. If AI is directed wisely, the ideal is not efficiency alone but the restoration of time and space for attentive listening, nuanced communication, and the recognition of distress with encounters for which human intelligence is best suited. Thus, a sustained research program into the human dimensions of ophthalmic care in the future AI world is warranted, equal in ambition to novel algorithm development. This should encompass areas such as the development and validation of screening instruments for diagnosis-related psychological distress, the creation and evaluation of interpretation and communication training frameworks tailored to ophthalmology’s unique clinical contexts, investigation of how clinician communication quality affects treatment adherence and study of how AI-liberated time can best be reinvested in patient-centered care [5, 6]. A cultural shift is also required within training programs, conferences, and journals toward recognizing that the human encounter is not a sentimental addendum to evidence-based medicine but is itself an evidence-based intervention, with measurable effects on outcomes that matter to patients [1]. If ophthalmology embraces this dual agenda, harnessing AI to liberate clinicians while investing with equal ambition in understanding and enhancing human interaction, it has an opportunity not merely to preserve the clinician’s role but to elevate it. The clinician of the future should not merely be a messenger of algorithmic outputs, but a more attentive, communicative, trusted, and compassionate practitioner, whose presence embodies what good care means.