Number
708
Name
Novel Application af Generative Artificial Intelligence-Assisted Problem-Based Learning: A Comparative Study in Medical Education
Date & Time
Monday, June 8, 2026, 6:00 PM - 7:30 PM
Location Name
Oglethorpe Ballroom
Authors
Christopher Matsko, NSU MD
Presentation Topic(s)
TBL/PBL
Description
PURPOSE
Generative artificial intelligence (GAI) shows growing promise in medical
education. Although ChatGPT can pass board examinations and offer students an
interactive learning environment, its role in problem-based learning (PBL) is
not well understood. This study evaluated the impact of integrating ChatGPT
into PBL sessions on learners’ clinical reasoning, communication, knowledge
acquisition, and team dynamics.
METHODS
Medical students were randomly assigned to facilitator-led or
ChatGPT-assisted PBL groups of 5–7 learners. All groups completed a
standardized PBL case. Data were collected through post-session knowledge
assessments and student perception surveys. Knowledge was measured as percent
correct, and perception items were grouped into four constructs: clinical
knowledge and skills, communication and collaboration, engagement and
affective response, and clinical reasoning and cognitive processing. Group
differences were analyzed using Welch’s t-tests.
RESULTS
Nineteen students participated in the ChatGPT-assisted group and 24 in the
facilitator-led group. There were no significant differences in knowledge
assessment scores (p=0.28). Perception scores were comparable for clinical knowledge
and skills (p=0.15), and communication and collaboration (p=0.33). Students
in the ChatGPT-assisted groups reported higher engagement and affective
response (p<0.05) and stronger perceptions of clinical reasoning and
cognitive processing (p<0.001), although the absolute magnitude of these
differences was modest.
CONCLUSION
Incorporating ChatGPT into PBL sessions resulted in higher student-reported
engagement and perceptions of enhanced clinical reasoning, without measurable
differences in knowledge acquisition or adverse effects on communication and
collaboration compared to traditional PBL. Our data suggest that
ChatGPT-assisted PBL sessions were not inferior to traditional
facilitator-led PBL, prompting the question of whether ChatGPT could, under
certain conditions, substitute for selected elements of human facilitation.
Answering this will require longitudinal studies across broader PBL
activities. Nonetheless, the present findings indicate that ChatGPT can be effectively
integrated into PBL within medical curricula.
Presentation Tag(s)
Student Presentation