SHM Converge Scientific Abstract ePoster Gallery-The Unsupervised Curriculum: Prevalence and Impact of Ai Tools Among Diverse Internal Medicine Residents in a Community Hospital

The Unsupervised Curriculum: Prevalence and Impact of Ai Tools Among Diverse Internal Medicine Residents in a Community Hospital

Back to course

Pdf Summary

This study surveyed Internal Medicine residents (PGY-2 to PGY-4) at Griffin Hospital to assess how AI tools, especially large language models, are being used in residency training and clinical work. It found that AI adoption is very high: 96% of residents reported using AI clinically. However, formal education is lacking, with only 18% receiving any structured AI training. Most learning occurs through self-study or peers, and 78% of residents support adding AI education to the curriculum.<br /><br />Residents most often used AI to expand differential diagnoses, summarize medical literature, and reduce documentation burden. AI scribe use was associated with an average 25% reduction in documentation time, freeing time for patient care. About 70% reported moderate improvement in diagnostic ability, suggesting AI is helpful but not transformative. Nearly half of respondents reported reduced burnout, while a similar proportion remained neutral.<br /><br />Despite heavy use, residents showed caution and low blind trust. They regularly verified AI outputs using guidelines and literature because of concerns about hallucinations, bias, accuracy, and privacy. Qualitative feedback suggested residents see AI as necessary for efficiency and “survival” in modern practice, but they also worry about overdependence and the possibility that increased efficiency could lead to higher workload expectations.<br /><br />Overall, AI has already become embedded in residency workflows, but its growth has outpaced formal Graduate Medical Education support. The authors argue that residency programs should develop structured AI curricula covering prompt engineering, model limitations, output verification, and safe clinical use.

Asset Subtitle

Arslan A. Khan

Meta Tag

Author List Arslan A. Khan, Sai Venkat Aluri, Siraj Memon

Category Research

Concept Internal Medicine Resident

Concept AI Adoption

Concept Residency AI Curriculum

Concept Prompt Engineering

Concept Output Verification

Distinguished Non-Finalist

Presenter Organization Griffin Hospital

Presenting Author Arslan A. Khan

Track Technology in Hospital Medicine

Keywords

Internal Medicine residents

artificial intelligence

large language models

residency training

clinical workflow

AI education

documentation burden

diagnostic support

burnout

prompt engineering

Internal Medicine Resident

AI Adoption

Residency AI Curriculum

Prompt Engineering

Output Verification