Related Experiment Videos
Artificial intelligence teaching assistants: a scalable solution for supporting struggling medical students
Alina Sami1, Mark Adkins1, Anne McLeod1,2
1MD Program, Temerty Faculty of Medicine, University of Toronto, Toronto, Ontario, Canada.
Summary
Large language models (LLMs) as AI-teaching assistants (AI-TAs) improved medical student exam performance and engagement. AI-TAs provided reliable support, enhancing psychological safety and aiding struggling students.
Area of Science:
- Medical Education Technology
- Artificial Intelligence in Healthcare
- Educational Psychology
Background:
- Large language models (LLMs) show promise in medical education tutoring.
- Previous studies focused on small-scale, targeted applications.
- The need for evaluating large-scale AI implementation in compulsory courses is evident.
Purpose of the Study:
- To assess the impact of large-scale AI-teaching assistants (AI-TAs) in a compulsory medical school course.
- To evaluate AI-TAs' effectiveness, usability, and influence on student learning.
- To determine if AI-TAs can support struggling students and complement traditional teaching.
Main Methods:
- A quasi-experimental observational study with mixed methods at the University of Toronto in 2024.
- AI-TAs were developed using OpenAI's ChatGPT-4o and implemented mid-course.
- Exam performance, surveys, and interviews were used to analyze impact and perceptions.
Main Results:
- Students using AI-TAs showed improved exam performance, converging with peers after adoption.
- AI-TA adoption reduced the proportion of students falling below the course assessment standard.
- Key advantages included reliable support, efficient application, and enhanced psychological safety.
Conclusions:
- AI-TAs correlate with improved exam scores and reduced academic difficulty.
- AI-TAs foster a psychologically safe environment, boosting active learning and engagement.
- AI-TAs represent a scalable, cost-effective tool for supporting medical students.
