Introduction

The integration of artificial intelligence (AI) into physician assistant (PA) curricula represents a timely priority as healthcare systems increasingly adopt AI-driven tools for diagnosis, treatment planning, documentation, and simulation-based education (Topol, 2019; Wartman & Combs, 2018). Guided by a technology acceptance–informed framework (Davis, 1989), this study examined PA students’ awareness, perceptions, and use of AI applications. A mixed-methods, cross-sectional online survey was distributed to Texas PA students to capture both quantitative trends and qualitative insights into student experiences.

Methods

A 15-item instrument was developed based on prior health education and technology adoption literature (Davis, 1989; Wartman & Combs, 2018). The instrument included demographic and program questions, familiarity with foundational AI concepts and tools (including large language models such as ChatGPT), frequency and context of AI use (e.g., study support, test preparation, and clinical simulation), perceived barriers and instructional preferences. Although the instrument was pilot tested for content clarity, formal psychometric validation was not established.

A total of 46 students from Texas-based accredited programs completed the survey, with more than 70% representing a single PA program that uses a hybrid instructional modality.

Quantitative data were analyzed descriptively. Differences in Likert-scale responses related to perceived confidence and familiarity with AI applications were compared between high-frequency AI users (daily or several times per week) and lower-frequency users using the Mann-Whitney U test.

Results

Findings indicated that the most commonly used tools included ChatGPT and other large language models, AI-based test preparation platforms, and clinical simulations. Over 41% of respondents reported using AI several times per week, while 37% reported daily use. High-frequency AI users had higher perceived confidence and familiarity with AI applications compared to lower-frequency users (Mann-Whitney U test, p < .01). Approximately 68.4% of students agreed that AI improved study efficiency and knowledge retention, and over 65% reported enhanced clinical reasoning. However, 55.3% expressed concerns regarding ethical and legal implications. Notably, only 36.8% supported formal curricular integration of AI, and just 26% felt adequately prepared to use AI in clinical practice.

Qualitative responses revealed themes of cautious optimism, uncertainty regarding ethical use, and a desire for structured guidance. While over 90% acknowledged the importance of AI literacy for future practice, students emphasized the need for clear instruction on validation of AI outputs, bias recognition, and responsible use.

Conclusions

These findings support the inclusion of concise, practice-oriented AI instruction within PA curricula, including modular learning, case-based applications, simulation experiences, and explicit training in critical appraisal and ethical considerations. Limitations include a small sample size, potential response bias, and limited generalizability due to geographic concentration and institutional representation.

Overall, this study highlights both enthusiasm and apprehension among PA students as AI becomes increasingly embedded in healthcare education and practice. The integration of artificial intelligence (AI) into physician assistant (PA) curricula represents a timely priority as healthcare systems increasingly adopt AI-driven tools for diagnosis, treatment planning, documentation, and simulation-based education (Gordon et al., 2024; Topol, 2019)

An example of an AI-related assignment utilized within our PA program involved students engaging with the BodySwaps platform (https://bodyswaps.co), an AI-powered immersive communication and soft skills training tool designed for healthcare education. The assignment demonstrates how AI-driven avatar simulations can support empathy, patient communication, teamwork, and clinical interaction skills in a psychologically safe learning environment. Students have remarked that the platform supports realistic interactions and opportunities for repetitive practice, yet expressed some concerns related to implementation within PA education. Future research should examine scalable curricular models, assessment strategies, and the impact of AI education on clinical performance and patient outcomes.

Acknowledgement

The authors would like to thank Dr. Jerrod Tynes, and Dr. Amy Bronson for their mentorship that inspired this work.