← Back to papers

How Does ChatGPT Perform on the United States Medical Licensing Examination (USMLE)? The Implications of Large Language Models for Medical Education and Knowledge Assessment

★ ★ ★ ★ ☆

Paper Summary

Paperzilla title
ChatGPT Passes Med School (Kinda): AI Aces Some Exams, Fails Others

ChatGPT demonstrated performance equivalent to a passing score for a third-year medical student on USMLE Step 1 and Step 2 practice questions, exceeding the accuracy of other large language models like GPT-3 and InstructGPT. While demonstrating logical reasoning in all responses and using internal information effectively, the model's reliance on external information was stronger for correct answers, highlighting a potential link between knowledge access and performance.

Explain Like I'm Five

Scientists found that a smart computer program called ChatGPT could pass really hard doctor exams, almost like a student who's been studying medicine for three years! It was even better than other computer programs.

Possible Conflicts of Interest

The authors acknowledge funding from the Yale School of Medicine and the National Institutes of Health, but declare no specific conflicts of interest.

Identified Limitations

Outdated training data
The study acknowledges that ChatGPT's training data is limited to information before 2021, potentially affecting its ability to answer questions about more recent medical advancements.
Limited access to model internals
The closed nature of the model and lack of public API prevented fine-tuning on task-specific data and a more thorough examination of its stochasticity.
Moving target problem
The rapid updates to ChatGPT introduce a moving target problem, meaning the model's performance could change significantly between evaluations.

Rating Explanation

This study provides a valuable early assessment of a large language model's capabilities in a critical domain, demonstrating promising results while acknowledging limitations. The methodology is sound, though constrained by the model's closed nature. No obvious attempts to manipulate the rating were detected.

Good to know

This is the Starter analysis. Paperzilla Pro fact-checks every citation, researches author backgrounds and funding sources, and uses advanced AI reasoning for more thorough insights.

Explore Pro →

Topic Hierarchy

Domain: Health Sciences
Field: Medicine
Subfield: Health Informatics

File Information

Original Title: How Does ChatGPT Perform on the United States Medical Licensing Examination (USMLE)? The Implications of Large Language Models for Medical Education and Knowledge Assessment
Uploaded: July 14, 2025 at 11:25 AM
Privacy: Public