Student Perceptions of Virtual Counseling Practice Using ChatGPT Voice Mode in Pharmacy Education

BACKGROUND: Large language model (LLM)-based tools such as ChatGPT offer potential to augment pharmacy training through scalable simulation. This study evaluated the acceptability and perceived utility of ChatGPT voice mode as a virtual patient for counseling practice among pharmacy students. METHODS: A cross-sectional post-intervention survey was conducted among fourth-year pharmacy students after completion of a ChatGPT-assisted virtual counseling practice activity using ChatGPT voice mode. A structured questionnaire assessed accessibility, realism, feedback quality, perceived professional skill support, and overall utility. The questionnaire underwent expert review for content validity using the item-objective congruence index. Descriptive statistics were used to summarize Likert-scale responses, and open-ended responses were explored using natural language processing techniques. RESULTS: Of 158 students, 137 (86.7%) completed the evaluation. Accessibility and feedback domains showed higher agreement, particularly regarding flexibility of practice and immediacy of feedback. Perceptions of realism and communication accuracy were more variable, with a substantial proportion of neutral and negative responses. Qualitative analysis identified key themes including practice flexibility, feedback utility, and limitations in conversational realism. CONCLUSION: ChatGPT voice mode may serve as a feasible and scalable adjunct for simulated counseling practice in pharmacy education. However, because this study evaluated students' perceptions of the learning experience rather than objectively measured competency, further research incorporating educator-rated performance and objective assessment of counseling skill is needed to determine its educational effectiveness.