#60 Teaching in the AI Era: Updates from AIMW26

June 11, 2026 | By Molly Heublein

Audio

With Drs Eric Burnett and Alexander Glaser

Explore how artificial intelligence is transforming medical education with insights from Drs. Eric Burnett (Columbia) and Alexander Glaser (Pennsylvania Hospital, UPenn), recorded live at AIMW26. This episode breaks down practical ways educators are using AI- from feedback tools and clinical reasoning support to EHR integration- while tackling real concerns like deskilling, bias, and academic integrity. Walk away with actionable frameworks and strategies to help learners use AI effectively, ethically, and in a way that actually improves their skills.

Claim CME for this episode at curbsiders.vcuhealth.org!

Website | Instagram | Twitter | Subscribe | Patreon | CME!| Youtube

 thecurbsidersteach@gmail.com

Meet our Guests:

Dr. Eric Burnett who is a hospitalist, Assistant Professor of Medicine, and IM Residency Associate Program Director at the Columbia University Irving Medical Center presenting on behalf of Drs. Rex Hermansen and Eve Merrill from their workshop “Artificial Intelligence Real Implications: Preparing Clinical Educators for the AI Era.” 

Dr. Alexander Glaser, Program Director of the IM Residency Program at Pennsylvania Hospital, generalist, and Assistant Professor of Medicine at the University of Pennsylvania Perelman School of Medicine presenting on behalf of himself and the planning committee for the AIM Week 26 precourse session they organized called “From Insight to Impact: Developing Yourself as a Next Generation Faculty Leader in AI.” 

Show Segments

• Intro, disclaimer, guest bio
• Guest one-liner/ Best piece of advice
• Benefits of AI
• Risks of AI
• Institutional Policies and Guidelines around using AI in MedEd
• Frameworks for Assessing AI Use
• Practical Considerations and Future Directions
• How to get started with AI as an Educator
• Balancing Climate/societal Concerns
• Outro

Artificial Intelligence in Medical Education Pearls

1. AI is already a part of training—don’t ignore it; engage learners openly to guide appropriate, skill-building use rather than driving it underground.
2. Use AI to augment, not replace thinking: great for feedback, case generation, and expanding differentials—but risky when it shortcuts clinical reasoning.
3. Watch for deskilling, mis-skilling, and “never learning” when AI replaces core cognitive work instead of supporting it.
4. Apply simple frameworks like pAIr (Purpose, Application, Integrity, Risk/benefit) to assess whether AI use is helping or harming learning.
5. Teach critical appraisal of AI outputs—verify sources, check for hallucinations, and recognize bias and “sycophancy.”

Artificial Intelligence in Medical Education Notes 

Our learners are using AI and will continue to do so, health professions educators need to be aware of these tools and work with our learners to optimize their use.

Benefits of Using AI in Medical Education

A 2025 NEJM article outlines three main benefits of AI in medical education: task offloading (e.g., summarizing narrative comments for fellowship letters or assisting with EHR tasks like billing), reducing cognitive load in patient care (e.g., AI/ambient scribes), and educational enhancement (Abdulnour 2025). AI can support medical education in many ways, including generating cases and providing targeted feedback on written histories and physicals. If you’re not sure where to start, consider your specific niche and tailor AI use to your needs.

Dr. Burnett finds AI particularly helpful in his work around remediation for clinical reasoning. Creating tailored cases aligned with specific learning goals can save time compared to sourcing real cases and allows educators to focus on targeted teaching points.

Dr. Glaser describes using AI to summarize narrative evaluations for fellowship letters—work that evolved into a plenary session at AIMW25.  He also highlighted sessions from the AIMW26 precourse, including Enhancing Learner Feedback with Actionable AI Tools by Drs. Daniel Sartori and Margaret Horlick.  They described how some programs, such as NYU, are using AI to enhance educational feedback through dashboards integrated into the EHR. These tools can scan resident notes, provide clinical insights, identify whether learners are seeing a sufficient volume of key diagnoses, and highlight missed diagnostic opportunities. This allows for real-time assessment and more actionable feedback.

Risks of Using AI

A healthy skepticism is warranted- like any new technology, AI is a double-edged sword.  If learners use AI to bypass practice or clinical reasoning, it can impair learning.  Overuse may lead to mis-skilling (reinforcing incorrect skills due to AI errors or bias), deskilling (loss of previously acquired skills), or even failure to develop foundational skills (Abdulnour 2025, Montieth 2026).  These risks span all levels of training; for example, even experienced endoscopists have demonstrated reduced independent polyp detection after relying on AI support (Budzyn 2025).

There is still clear value in doing core tasks manually- such as writing notes or building differential diagnoses- to develop clinical skills.  AI is most beneficial when used with intention (e.g., creating study guides, generating practice cases, or receiving feedback), and less so when it replaces effort entirely.

Additional risks include broader societal and environmental impacts, as well as concerns about academic integrity and HIPAA compliance.  For example, if a learner submits fully AI-generated work, questions about authorship and plagiarism arise.

Educators are still determining the best timing for introducing AI into training. Should learners demonstrate competency in core skills before using these tools, or should AI be incorporated early, given its likely role in future practice?  Work from Johns Hopkins/Bayview presented by Drs. Janie Abernethy, Paul David O’Rourke, and Amteshwar Singh at the AIMW26 precourse titled Ensuring Learner Readiness for Patient Facing AI Tools highlighted the importance of assessing learner skills before implementing AI supports such as ambient scribes and ensuring equitable access and training.

Policies and Guidelines Around AI Use

Institutional policies on AI use in education vary widely, and many programs lack clear guidance. Existing policies often emphasize restrictions rather than offering practical direction on appropriate use. More effective frameworks would include specific guidance on how to use approved, HIPAA-compliant tools (Triola 2025). Institutional support can help normalize AI use, particularly as faculty adoption varies. However, the rapid evolution of AI makes maintaining up-to-date policies challenging.

Dr. Luise Daniel Lugo in his AIMW26 session Now Make it Legal: AI Ethics, Policy, and Privacy in GME emphasized the importance of privacy, ethics, and legal considerations in AI use in GME.  Dr. Glaser recommends collaborating with institutional leaders (e.g., Chief  Medical Officers, Chief Informatics Officers, informatics, and legal teams) to develop policies. For those interested, there is an opportunity to serve as local champions- bringing in experts, organizing workshops, and advancing education in this space.

Frameworks for Assessing AI Use

Dr. Burnett highlighted the pAIr framework (developed by Dr. Rex Hermansen and presented in their AIMW26 session):

• purpose: Why is AI being used?
• Application: How is it being used?
• Integrity: Does it support professional identity and is it being used ethically?
• risk/benefit: Does it enhance or undermine learning?

This framework can be applied in real time—for example, when having a learner reflect on using AI to expand a differential diagnosis as an appropriate use.  Importantly, educators should avoid “AI shaming,” as learners may otherwise hide their use of AI, making it harder to identify knowledge gaps.

Other frameworks, such as the NEJM DEFT-AI approach (𝐃𝐢𝐚𝐠𝐧𝐨𝐬𝐢𝐬, 𝐄𝐯𝐢𝐝𝐞𝐧𝐜𝐞, 𝐅𝐞𝐞𝐝𝐛𝐚𝐜𝐤, 𝐚𝐧𝐝 𝐓𝐞𝐚𝐜𝐡𝐢𝐧𝐠) 𝐅𝐫𝐚𝐦𝐞𝐰𝐨𝐫𝐤 𝐭𝐨 𝐏𝐫𝐨𝐦𝐨𝐭𝐞 𝐂𝐫𝐢𝐭𝐢𝐜𝐚𝐥 𝐓𝐡𝐢𝐧𝐤𝐢𝐧𝐠 𝐝𝐮𝐫𝐢𝐧𝐠 𝐚𝐧 𝐀𝐈 𝐈𝐧𝐭𝐞𝐫𝐚𝐜𝐭𝐢𝐨𝐧), can also help assess how learners integrate AI into their thinking (Abdulnour 2025).

Practical Considerations and Future Directions

AI has the potential to improve clinician-computer interfaces and streamline EHR use, allowing clinicians to refocus on bedside patient care. However, generative AI outputs can vary, so learners must be taught to critically appraise results- verify sources, confirm cited studies, and ensure recommendations apply to the patient.  As Dr. Burnett notes, AI can exhibit “sycophancy,” tending to agree with users, which reinforces the need for critical evaluation.  Faculty development is essential to model effective and ethical AI use. 

Dr. Glaser highlights an example shared at the precourse where learners watch a pre-recorded video of a patient history-taking and then are asked to write up a history of present illness (HPI). They then share the HPI that an AI scribe created to help learners compare to their own to see the pitfalls and skills of these tools. 

Dr. Glaser also highlights evolving educational priorities- for example, if AI reliably interprets EKGs, training may shift toward development of other skills.  

How to Get Started

For those new to AI, the best approach is to experiment.  Dr. Burnett suggests inputting one of your talks into AI and asking how to improve it. Or ask a chatbot to give a talk on the same topic, and compare if there’s anything you’ve left out or if there’s a different way to structure your talk. Consider asking AI to tailor your talk to different levels of learners, for example a third year medical student vs. a fellow, to see what you might adjust or augment in your lecture. Play with using OpenEvidence to write insurance appeal letters.  Review AI-generated summaries in your EHR, if available, to understand their strengths and limitations. Experienced clinicians may be better equipped to identify gaps that learners might miss.

Societal and Environmental Considerations

Some learners may have concerns about the environmental or societal impact of AI. While these concerns are valid, AI is becoming increasingly integrated into healthcare, making complete avoidance unrealistic.  While we should not ignore/minimize learners’ valid concerns about these, placing it in perspective that healthcare already has a large environmental impact, and just as we should be choosing wisely around over-ordering diagnostic tests, we should be choosing wisely about how we are using AI in a responsible way.

Links

1. Abdulnour RE, Gin B, Boscardin CK. Educational Strategies for Clinical Supervision of Artificial Intelligence Use. N Engl J Med. 2025 Aug 21;393(8):786-797. doi: 10.1056/NEJMra2503232. PMID: 40834302.
2. Dr Burnett highlights his work as a digital media fellow with the Journal of Hospital Medicine’s visual abstracts 
3. Dr Glaser is on the Program Planning Committee for APDIM and encourages attendance at the Fall 2026 APDIM meeting

Goal

Listeners will recognize the benefits that artificial intelligence brings to health professions education and how to prepare for the AI Era as clinician-educators.

Learning objectives

After listening to this episode listeners will be able to …

1. Review the benefits and risks of the use of artificial intelligence for health professions education
2. Name some examples of how AI tools can be used in the clinical learning environment
3. Discuss guidance for responsible, ethical, and professional use of AI for education

Disclosures

Drs. Burnett and Glaser report no relevant financial disclosures. The Curbsiders report no relevant financial disclosures. 

Citation

Heublein M, Cheng MKW, Burnett E, Glaser A, Kryzhanovskaya E. “#60 Teaching in the AI Era: Updates from AIMW26.  The Curbsiders Teach Podcast. https://thecurbsiders.com/teach June 11, 2026.