Big Data/Machine Learning/AI
Understanding AI use in epidemiology education: attitudes, predictors, and implications Teresa M. Janevic* John B. Wetmore Natalie A. Boychuk Marco Thimm-Kaiser Teresa M. Janevic
Artificial intelligence (AI) is transforming education, yet little is known about AI use among epidemiology students or its impact on learning.
We anonymously surveyed 109 master’s students enrolled in a coding-intensive epidemiology course based on the Modern Epidemiology textbook. We used multinomial logistic regression to estimate the relationship between attitudes toward AI, measured by the validated AI Attitude Scale (AIAS), and AI use frequency. We also used linear regression to estimate the relationship between AI use frequency and perceived learning gains, measured by the Student Assessment of Their Learning Gains (SALG) instrument. Both models adjusted for age, sex, race-ethnicity, language spoken at home, and nativity.
Among students, 28% used AI infrequently (“never/rarely”), 39% occasionally (“sometimes”), and 33% frequently (“often/always”). ChatGPT was the most used resource (84%), followed by Grammarly (17%). Students primarily used AI to understand concepts (71%), write SAS code (49%), and improve writing quality (40%). Regression modeling (50%), survival analysis (47%), sources of bias (44%), and interaction (44%) were the most common topics for which AI was used. Students largely regarded AI as helpful (67%).
For every one-unit increase in a student’s AIAS score, their odds of frequent compared to infrequent AI use increased by 80% (OR=1.8, 95% CI: 1.4–2.4); however, attitudes did not predict occasional use (OR=1.1, 95% CI: 0.9–1.4) after adjustment. When comparing infrequent and occasional AI users to frequent users, there was no meaningful difference in perceived learning gains (β=0.1, 95% CI: -0.1–0.2).
Positive attitudes toward AI predict frequent use, but AI use frequency does not correlate with perceived learning gains. These findings highlight the frequency and types of AI use among epidemiology graduate students. Future research should explore applications of AI that optimize student learning and achievement.