JMIR Medical Education

Clinical summary statements convert unstructured patient information into a concise format. They are widely used in virtual patients (VPs) to train medical students in diagnostic reasoning. Most VP systems provide little individualized feedback on the quality of learners’ summaries, and earlier automated approaches have required complex natural language processing, limiting their scalability. Large language models (LLMs) may offer a more flexible, rubric-guided way to score summary statements and support feedback. Still, their test-retest reliability and concordance with expert ratings in this context remain unclear.

Student engagement reforms have increased active learning in higher education. However, visible participation does not always require students to rehearse the spoken, interactive forms of reasoning central to professional practice, such as explaining one’s thinking, responding to challenges, building on others’ ideas, and working through uncertainty in real time. We describe this gap as an engagement blind spot. We define audible reasoning as the deliberate externalization of disciplinary or professional reasoning through spoken interaction so that its assumptions, evidence, uncertainty, and justification can be examined, challenged, revised, and developed collaboratively. It is particularly relevant in health professions education and also applies to laboratory-based and scientific disciplines where students must communicate their reasoning under conditions of uncertainty, scrutiny, and collaboration. Digital, hybrid, asynchronous, and AI-enabled learning environments sharpen this blind spot. Shared documents, online quizzes, discussion boards, and generative AI tools can support participation and access; however, they can also allow students to engage without speaking or to produce polished answers without making the underlying reasoning visible. In this context, educators need to ask not only whether students are engaged but also what behaviors they are rehearsing and whether their reasoning is available for examination and development. This tutorial offers 12 practical, evidence-informed tips developed iteratively from recurring challenges encountered in teaching, curriculum design, and faculty development and refined against selected literature. The tips are organized around 4 clusters covering the design of activities that require audible reasoning, the elicitation of reasoning in practice, the development of reasoning through listening and response, and the preservation of audible reasoning in digital and AI-enhanced environments. Across these 4 clusters, we suggest small but intentional shifts that help educators move from silent, constructive tasks toward audible, dialogic, and professionally relevant participation. These include designing tasks around uncertainty, creating low-stakes opportunities for spoken reasoning, using digital tools to surface and organize contributions for further discussion, structuring active listening and peer feedback, reviewing group dialogue formatively, and using conversational AI as a rehearsal partner rather than an authority. These strategies require attention to psychological safety, cultural responsiveness, accessibility, equity, privacy, data governance, and ongoing monitoring. Audible reasoning is not synonymous with speaking more, volunteering more frequently, or sounding fluent. Designing for audible reasoning is not simply an engagement strategy. It is a way of helping students practice the communicative and collaborative behaviors needed to explain, defend, question, and revise human judgment in clinical, scientific, and other professional contexts.

Serious illness communication (SIC) is a critical skill in medical education, especially in end-of-life care, yet clinicians often lack sufficient opportunities for deliberate practice. Standardized patients (SPs) remain the gold standard for SIC training, but they are limited by cost, availability, and consistency. To support wider access to SIC training, scalable alternatives are needed.

Generative AI, including large language models (LLMs), may enable personalized learning in medical education. In otolaryngology–head and neck surgery (ear, nose, and throat [ENT]) education, AI may support knowledge retrieval, adaptive assessment, feedback, and scalable learning, but evidence remains fragmented.

Urology education must prepare learners to integrate anatomical visualization, diagnostic reasoning, technical performance, communication, and ethical judgment, often under restricted working hours and limited access to high-fidelity cases. This viewpoint argues that large language models (LLMs) and metaverse-related technologies should be integrated as complementary, supervised educational tools rather than treated as autonomous replacements for educators or clinical experience. LLMs can support cognitive apprenticeship, documentation practice, assessment, and communication simulations, whereas extended reality, cloud computing, and virtual digital humans can support spatial, procedural, and remote learning. Our aim is to articulate 3 key messages: first, the educational value of these technologies depends on alignment with defined urological competencies; second, implementation should proceed in stages from governed data infrastructure to intelligent interaction and immersive multimodal training; and third, educational effectiveness must be established through prospective evaluations of transfer to clinical performance, not through technical novelty alone. We emphasize the need for rigorous randomized controlled trials, identify requirements for haptic and biomechanical fidelity, and discuss hallucination, bias, privacy, accountability, and deskilling. The responsible future of intelligent urology education will therefore require collaboration among learners, educators, clinicians, engineers, ethicists, institutions, and professional bodies.

Artificial intelligence is rapidly reshaping clinical research, diagnostics, health system operations, and medical education. Frontier generative AI development is increasingly concentrated in private industry, whose computational infrastructure, proprietary datasets, and capital exceed those of most academic institutions. Although not designed exclusively for medicine, these systems increasingly influence clinical applications. This shift raises a question for academic medical centers (AMCs): What role remains when leading AI systems emerge outside universities? This viewpoint examines industry’s growing role in generative AI, clarifies AMCs’ responsibilities, and proposes a framework for AI governance and literacy across undergraduate, graduate, and continuing medical education (CME). Rather than compete with industry on computational scale, AMCs should reinforce functions for which AMCs have distinctive strengths. These include (1) mechanistic inquiry linking algorithmic outputs to biological and pathophysiologic understanding; (2) interdisciplinary synthesis integrating clinical, ethical, legal, and sociotechnical perspectives; (3) independent and methodologically rigorous validation of clinical AI tools; and (4) institutional stewardship of ethical governance for patient populations that are often underrepresented in commercial datasets. As nonprofit institutions training clinicians, AMCs have a distinct role in safeguarding transparency, fairness, and accountability in AI-enabled health care. We position this 4-role framework to refine AI governance and translational medicine approaches, clarifying which responsibilities AMCs should retain while collaborating with industry. The approach includes longitudinal AI literacy training across undergraduate medical education, graduate medical education, and CME, standardized pathways for evaluating clinical AI, stronger conflict-of-interest safeguards, and complementary partnerships with industry and regulators. We also outline educational assessment methods and safety-monitoring processes, including competency-based assessment across undergraduate medical education, graduate medical education, and CME, postdeployment monitoring, and clear escalation triggers. Embedding verification habits and governance into training and practice can help ensure that generative AI supports clinical reasoning while preserving trust, equity, and scientific rigor. This viewpoint addresses academic leaders, curriculum committees, residency directors, faculty developers, and accreditors. Safe clinical AI requires sustained academic leadership.

Simulation-based education is increasingly used in medical training to safely develop clinical skills. However, the perceived relevance of a simulation (“utility value”) to a learner’s future work may influence its effectiveness.

The undergraduate stage is critical for developing nursing students’ disaster nursing competencies; however, traditional lecture-based teaching often fails to promote active learning, and innovative instructional models remain underused in this field. The BOPPPS (Bridge-in, Objective, Preassessment, Participatory Learning, Postassessment, and Summary) model is a student-centered, closed-loop instructional framework that emphasizes active participatory learning and real-time feedback.

Competency-based medical education increasingly uses digital workplace assessment platforms to collect longitudinal evidence of trainees’ readiness for progressive responsibility. Entrustable professional activities (EPAs) translate competencies into observable clinical work, but individual workplace observations are context-dependent and may not support dependable progression decisions. Empirical guidance remains limited on how many observations, faculty raters, EPA titles, and workplace settings are needed for a dependable EPA portfolio.

Continuing medical education (CME) and continuing professional development (CPD) systems have traditionally relied on time-based credit allocation, using participation duration as a proxy for professional learning. Although administratively simple and scalable, this model does not reliably demonstrate whether physicians have engaged in meaningful learning, improved clinical reasoning, or critically appraised evidence. The emergence of generative AI creates an opportunity to rethink how physician learning is documented, assessed, and credited. This viewpoint proposes the Case-based Learning Intelligence Credit System (CLICS), a conceptual framework for translating AI-mediated clinical learning interactions into auditable evidence of reasoning-related engagement that could support CME/CPD credit. CLICS introduces the professional learning episode (PLE) as the basic unit of creditable learning: a coherent AI-mediated interaction demonstrating a clinically meaningful problem, reasoning development through iterative inquiry, contextual or evidentiary integration, and reflective synthesis. PLEs are evaluated using the proposed Practice Intelligence Score-7 (PIS-7) rubric, subject to human calibration and oversight; the rubric assesses observable reasoning behavior within the episode rather than the AI’s answer, and qualifying PLEs may be translated into CME/CPD credit through threshold-based, human-auditable conversion rules. CLICS is not intended to replace traditional CME but to extend it as an optional, evidence-generating pathway for personalized, practice-embedded professional development. Its implementation requires iterative validation, stratified human audit, privacy-by-design architecture, antigaming controls, bias monitoring, and professional oversight. If validated, CLICS may enable identification of domain-specific areas for improvement and support personalized, adaptive learning pathways.

The integration of AI into health care is increasingly shaping clinical practice and decision-making. Beyond technical performance, AI has implications for professional roles, clinical reasoning, and responsibility. Understanding how medical students and clinicians perceive AI, how these perceptions relate to professional identity concerns, and how clinicians evaluate AI-supported decisions in different clinical contexts is therefore important for both implementation and medical education.

Early discussions of generative language models in medical education emphasized their promise for simulation, digital patients, individualized feedback, learner assessment, health information dissemination, research support, and translation, while also warning about bias, privacy, academic integrity, misinformation, legal ambiguity, and unequal access. Since that first wave, generative AI has moved from novelty to routine exposure for learners, educators, researchers, and institutions. Medical education therefore needs a more mature framework than a catalog of opportunities and risks. This viewpoint argues that the next phase should be organized around educational entrustment: determining which functions can be delegated to AI systems, under what conditions, with what human supervision, and with what evidence of benefit. Building on recent proposals to apply entrustment to AI in health professions education, we operationalize the concept into a graduated, function-level model that specifies which educational functions may be delegated; at what stakes; and with what oversight, assessment, and governance. We classify use cases by educational stakes and AI autonomy and outline implications for assessment redesign, curriculum development, faculty capability, cognitive autonomy, equity, and institutional governance. The central challenge is whether medical schools can integrate these tools in ways that preserve clinical reasoning, professional identity, accountability, and fairness. The next generation of research should move beyond model performance on examinations and evaluate how AI changes learning, judgment, behavior, and patient care.