Current Landscape of Generative AI Use as a Search Engine Among Resident Physicians: Cross-Sectional Study

Area of Science:

• Medical Education
• Artificial Intelligence in Healthcare
• Clinical Practice

Background:

• Generative AI (GenAI) literacy is crucial for resident physicians to prevent knowledge collapse and skill deficits in the evolving healthcare landscape.
• Resident physicians are highly susceptible to GenAI's influence, including those who are not yet users.

Purpose of the Study:

• To examine the extent of GenAI use as a clinical search engine among resident physicians.
• To compare GenAI literacy and bedside behaviors in infectious disease management between GenAI users and nonusers.

Main Methods:

• A cross-sectional survey was administered to resident physicians participating in Japan's General Medicine In-Training Examination (GM-ITE).
• The survey assessed GenAI usage, GenAI literacy, and perceived bedside behaviors, alongside demographic data and GM-ITE scores.
• Statistical analyses compared attitudes and behaviors between GenAI users and nonusers, adjusting for relevant covariates.

Main Results:

• Of 2850 respondents, 39.4% used GenAI as a clinical search engine; only 13% used it for primary differential diagnoses.
• GenAI users demonstrated higher understanding of GenAI limitations (e.g., confabulation) and essential competencies (e.g., history taking, physical examination) than nonusers.
• GenAI use was associated with improved perceived behaviors, including thorough examination and appropriate antimicrobial use, and greater awareness of ethical considerations like fairness and transparency.

Conclusions:

• GenAI use did not correlate with self-reported knowledge collapse indicators at this stage.
• GenAI literacy is limited among resident physicians, especially nonusers, highlighting a need for targeted education.
• Education on GenAI in clinical practice, focusing on critical thinking and ethics, is recommended for both users and nonusers.