Generative Artificial Intelligence Use in United States Doctor of Physical Therapy Programs
Craig P Hensley, Christa M Nelson, Benjamin Price, Allison Dennemann, Benjamin Knoll, Carolyn Oh, Jungwha Lee, Carole A Tucker, Alison H ChangAbstract
Importance
Generative Artificial Intelligence (AI) in education is poised to significantly impact the field of physical therapy. Research on generative AI use in physical therapist education remains limited.
Objective
The objective was to examine generative AI use in physical therapist education among students and faculty.
Design
This was a cross-sectional, online survey.
Setting
The survey was conducted among accredited Doctor of Physical Therapy (DPT) programs in the United States (US).
Participants
Students and faculty were recruited from US DPT programs as of January 2025.
Intervention(s) or Exposure(s)
Generative AI use within DPT programs was measured by participant self-report.
Main Outcomes and Measures
A novel 32-item survey comprising closed- and open-ended questions assessed generative AI use, patterns and purposes of use; platforms, training and comfort levels, views on generative AI’s role in physical therapist practice, student preparedness and academic integrity, adoption facilitators/barriers, and demographics. Factors associated with generative AI use were examined using multivariable logistic regression.
Results
Among 1067 respondents (873 students, 194 faculty), 853 (79.9%) reported using generative AI, including 687 students (78.7%) and 166 faculty (85.6%). ChatGPT was the most common used platform (662 students, 96.4%; 153 faculty, 92.2%). Common reasons for use included creating study guide/sample questions (482 students, 70.2%; 74 faculty, 44.6%) and creating patient cases (370 students, 53.9%; 102 faculty, 61.5%). Barriers to use included concerns about reliability and scientific rigor (student users: 411, 59.8%; non-users: 133, 71.5%; faculty users: 90, 54.2%; non-users: 14, 50.0%) and lack of trust in the technology and/or companies behind it (student users: 287, 41.8%; non-users: 112, 60.2%; faculty users: 63, 38.0%; non-users: 13, 46.4%). Generative AI use was associated with higher self-reported comfort with the technology (students: OR = 11.6, 95% CI = 7.9–17.1; faculty: OR = 14.3, 95% CI = 5.1–40.0).
Conclusion
Generative AI is widely used among DPT students and faculty, yet substantial variability exists in its use and formal instruction within DPT programs.
Relevance
As its integration into health care accelerates, incorporating education on generative AI within DPT programs is crucial to ensure students are prepared to use these tools competently and responsibly.