Presented By: John Lee, University of Arkansas for Medical Sciences
Co-Authors: David Davies, University of Arkansas for Medical Sciences
Evan Hicks, University of Arkansas for Medical Sciences
Tiffany Huitt, University of Arkansas for Medical Sciences
Erica Malone, University of Arkansas for Medical Sciences
John Norys, University of Arkansas for Medical Sciences
Humam Shahare, University of Arkansas for Medical Sciences
Vanessa Weidling, University of Arkansas for Medical Sciences

Purpose
One of the challenges for medical students is developing skills for self-directed learning. The rapid pace at which content is presented limits the ability to prepare for lectures in advance. The utilization of Generative Artificial intelligence (AI) is growing rapidly in many industries due to its potential for increasing productivity. However, the utility of AI as a tool for self-directed learning in medical education is not well defined. The purpose of this project is to determine if using generative AI, ChatGPT, is an effective augment for self-directed learning and lecture preparation for first year medical students in an Anatomy course. 

Methods
In this cross over design, first year medical students (n=32) independently studied new anatomy content in advance of a faculty lecture using provided resources under two conditions: with and without ChatGPT as a resource. All students participated in both conditions (ChatGPT, non-ChatGPT) and completed pre-and post-assessments after each session of independent study to measure efficiency in learning. Student performance scores were compared between pre-post assessments and between conditions. 

Results
In both conditions, there was no difference among students in baseline knowledge of content (p =0.85350). In the ChatGPT condition, post assessment scores were significantly higher than baseline scores (p =0.00293). Importantly, this was not observed for the non-ChatGPT condition (p=0.21371). In addition, ChatGPT post assessment scores were significantly higher than non-ChatGPT scores (p =0.02731), indicating improved self-directed learning. 

Conclusion 
We conclude ChatGPT can be utilized to augment the self-directed learning experience of medical students when preparing for anatomy lectures. Even with limited experience using ChatGPT, our study shows this technology is capable of directing the learning of medical students. It is imperative to educate medical students on the way in which this technology can benefit their learning process in this era of massive medical and technological growth.