CBME in the AI Era:
Challenges, Opportunities, and Directions

ABSTRACT

Recent Artificial intelligence advance is transforming the skills physicians need and reshaping competency-based medical education (CBME). Beyond clinical reasoning, communication, and proficiency, future physicians must master AI literacy, data interpretation, ethical oversight, and human-AI collaboration. This session discusses a framework that integrates critical assessment of algorithmic outputs and adaptive expertise into CBME. Strategies include case-based learning, interdisciplinary team work with data scientists, and reflective real-world practice on when to trust or override AI. Teaching colloquiums and assessment methods, including entrustable professional activities and workplace evaluations such as interactive workshops, ethical case discussions, and simulated clinical scenarios will be presented. By aligning CBME with AI-enabled competency learning, educators can prepare physicians to use technology responsibly while ensuring patient-centered, equitable, and high-quality practice to further improve patient care.

ABSTRACT

Precision medical education integrates clinical and educational data to support more continuous, individualized, and outcome-informed learning. In the AI era, this approach is becoming increasingly feasible across the continuum of medical education. This talk will present a practical framework for precision medical education, with examples of AI-enabled assessment, feedback, coaching, and workplace learning. It will also highlight the infrastructure needed to support implementation, including data integration, analytic capacity, governance, and faculty engagement. Participants will leave with a clearer understanding of how institutions can begin building systems that learn from routine practice and better support learner development over time.

ABSTRACT

Education programs routinely gather and synthesize trainee data to improve learning. Over the past decade, the shift toward CBME has prompted better integration of assessment systems with data science, incorporating learning analytics and careful consideration for implementation strategies. The emergence of artificial intelligence (AI) has transformed the value of assessments in CBME that inform patterns of developmental trajectories (learning curves), progress toward competence, and alignment of training performance with healthcare outcomes. This presentation will discuss innovations in data science and AI using data collected through local and national initiatives, with implications for methodologies, delivery of CBME, and effective educational implementation.

ABSTRACT

Artificial intelligence (AI) is increasingly used in health professions education, but its role raises important questions about trust and entrustment. This presentation distinguishes entrusting learners with AI from entrusting learners to AI, arguing that AI cannot participate in reciprocal human trust because it does not bear responsibility, relational risk, or shared vulnerability. Instead, AI should be understood as a technological lens that can mediate trust between patients, trainees, supervisors, and educational systems. Drawing on trust theory, workplace-based assessment, and examples from AI-assisted data extraction and virtual patient assessment, the presentation proposes a task- and context-specific approach to determining appropriate AI autonomy. It also introduces a staged safety blueprint, progressing from AI-AI testing to expert testing, learner volunteer testing, and eventual deployment. Safe use of AI in education requires attention to stakes, context, human oversight, and the effects of AI on human relationships.

From Center to Department: Pre-positioning the Institutional Scaffolding for AI-Era CBME -- Chia-Te Liao, MD, PhD

Embedding generative AI into Competency-Based Medical Education (CBME) at scale requires more than tool deployment—it demands deliberate institutional pre-positioning. This session draws on Chi Mei Medical Center's three-year transformation, from the 2023 "Four Anchors of Governance" (governance architecture, learning pathway, faculty development, and AI engineering) to the 2025 ten-year framework, illustrating how an "organization-first, tools-after" strategy created a hospital-wide educational ecosystem capable of absorbing AI.
Two pre-positioning moves preceded any AI deployment. First, the Medical Education Center was elevated to a Medical Education Department, granting teaching leadership the institutional standing to embed AI governance into faculty development, EPA design, and assessment. Second, a four-pillar, six-domain architecture was established under the principle of "AI empowerment with humanistic foundation," forming receptive scaffolding for AI integration. The BRIDGE framework systematized the three CBME solutions—visualizing workplace learning, dissolving assessment silos, and aligning educator cognition—creating the surface onto which AI could be coupled.
Building on this scaffolding, the A+ Holistic Medical Education System (four modules) and HIS Copilot were deployed as an integrated rollout rather than standalone tools, ultimately serving over 160,000 monthly active users and earning the Edison Award 2025, iF Design Award 2026, HBR Taiwan Ding-Ge Award, and IHF Global Top-3 recognition.
The talk closes with three "if-not" principles—without organizational standing, without receptive scaffolding, and without sustainability design, AI-era CBME transformation cannot endure. Drawing on the closing of the first five-year "Carp's Leap" plan and the looming next "Dragon's Ascent" decade, the session distills transferable pre-positioning principles for fellow medical education leaders preparing for the next wave.

When Clinical AI Enters the Real World: Reflections on Competence, Trust, and Medical Training -- Shih-Sheng Chang, MD, PhD

As clinical AI moves beyond research publications and demonstration projects into real-world healthcare settings, it brings not only opportunities for greater efficiency and workflow optimization, but also new challenges to our understanding of professional competence, clinical judgment, and trust in healthcare.
Drawing on firsthand experience implementing AI in clinical practice, this presentation will share practical examples from cardiovascular medicine, medical imaging interpretation, clinical documentation and summarization, decision support systems, and smart hospital workflows. Through these experiences, we will explore how the integration of AI is reshaping the capabilities and mindsets required of physicians and healthcare teams.
Using real-world cases and practical reflections, this session invites medical educators to consider a fundamental question: AI is not merely a new tool—it is also a mirror that compels us to re-examine the core competencies that medical training should cultivate. As AI becomes increasingly embedded in everyday clinical practice, the task of medical education extends beyond teaching learners how to use AI. It must also prepare them to exercise sound judgment, build trust, and continue learning in AI-enabled environments, ultimately becoming healthcare professionals worthy of the trust of patients and society.

The AI-Augmented Competency Model: Redefining Medical Education in the Era of Human-AI Symbiosis -- Cheng-Heng Liu, MD

The rapid integration of Generative AI into healthcare has fundamentally transformed clinical workflows and academic research. However, this paradigm shift brings unprecedented challenges to medical education, particularly the risks of "cognitive surrender" and "de-skilling" among young physicians and trainees. As AI systems become more capable, the traditional competency-based medical education (CBME) frameworks must evolve.
This keynote introduces the AI-Augmented Competency Model, a novel framework designed to prepare medical professionals for the era of human-AI symbiosis. The presentation will explore its three foundational pillars: Dual Literacy (the intersection of domain expertise and AI literacy), Shared Control (mitigating automation bias and retaining high-level clinical decision-making), and the Joint Workspace (moving beyond basic prompting to integrate AI as a collaborative partner in continuous clinical and academic environments). By redefining what it means to be a "competent" physician, this session provides educators and clinical specialists with a pragmatic roadmap to foster the next generation of AI-resilient medical professionals.

Generative AI in Medical Education: Where Can It Actually Help? -- Wei-Chun Wang, MD

The rapid advancement of generative AI has prompted growing interest in its potential applications across healthcare and education. Yet in medical education, many discussions remain at the level of possibility rather than practice—leaving clinician-educators with an important unanswered question: where can generative AI actually help?
The applications we discuss span several dimensions of medical education: simulation-based training, clinical documentation and feedback, knowledge support at the point of care, and clinical reasoning. For each, we reflect on what the technology can genuinely contribute, where it falls short, and what conditions are necessary for it to be useful rather than merely impressive. Our aim is not to advocate for AI adoption, but to offer an honest, experience-based perspective that helps educators and institutions ask better questions—and make more grounded decisions—about where generative AI fits in the future of medical education.

ABSTRACT

Artificial intelligence (AI) is currently disrupting healthcare and medical education. Recent and rapid advances in AI continue to accelerate, appearing to outpace our ability to adapt and integrate AI into our daily work. The challenge before us is how to most effectively, and ethically, use these new AI technologies to achieve the aims of competency-based medical education (CBME) and its focus on ensuring physicians are fully prepared to provide high quality health care. While some AI research shows substantial promise to improve the effectiveness of clinical care and the professional development of health professionals, other studies report mixed results. Substantial uncertainty remains regarding the optimal role and use of AI. This is especially true for competency-based medical education programs where the ultimate desired educational outcome is mastery. This brief talk will explore how AI can better outcomes in complex medical education ecosystems, and the implications for the design and implementation of CBME programs.