Artificial Intelligence in Shared Decision-Making Education: A Scoping Review with Ethical Implications for Designing Virtual Standardized Patients

Abstract

Shared decision-making (SDM) operationalizes respect for patient autonomy, yet health professions education often evaluates it solely through self-reported confidence. This scoping review synthesized how artificial intelligence (AI) is used to teach, assess, or support SDM competency among clinical trainees and practitioners. A search of four databases in August 2026 identified 1,008 unique records, with 13 meeting inclusion criteria. AI served as an instructional environment (n=5), an assessment infrastructure (n=6), or catalyst for curricular improvement (n=2). Only one randomized educational trial was identified, and no study demonstrated durable transfer to clinical practice. Eleven sources explicitly targeted patients’ values and preferences, yet only one addressed ethical reasoning, and only two of five simulated encounter models articulated clinical options. Current approaches therefore emphasize the informational dimension of consent at the expense of collaborative deliberation. Accordingly, we propose five ethical design principles for virtual standardized patients, formulated as a design specification warranting empirical evaluation.