AN EXPERIENTIAL STUDY ON THE APPLICATION OF ARTIFICIAL INTELLIGENCE IN PSYCHOLOGICAL COUNSELING TECHNIQUE TRAINING FOR COLLEGE STUDENTS

Abstract

Objective: To investigate clinical medicine students’ experiences with, evaluations of, and perceptions of the application boundaries of generative artificial intelligence (AI) in assisting psychological counseling technique training, with particular attention to listening response techniques. Methods: A self-designed Questionnaire on Experiences with AI-Assisted Psychological Counseling Technique Training was administered to 118 clinical medicine students at a university. Descriptive statistics and frequency analysis were used to examine students’ practice patterns, interaction modes, perceived usefulness of AI-generated feedback, perceived improvement in counseling competence, overall satisfaction, evaluations of different AI models, and perceptions of the aspects in which AI could not replace human-led training. Results: The most common practice frequency was one to two sessions per month (66.1%), while the most common interaction mode was simulated dialogue through role-playing a client (41.5%). The mean scores for the helpfulness of AI feedback in understanding listening techniques, perceived improvement in counseling competence, and overall satisfaction were 4.19 (SD = 0.73), 4.01 (SD = 0.86), and 4.25 (SD = 0.75), respectively. DeepSeek received the highest proportion of favorable evaluations in response professionalism (61.0%), feedback detail (54.2%), and training suitability (50.8%), and was also selected most frequently as the overall model considered suitable for assisted training (55.9%). The main areas in which students believed AI could not replace human-led training were emotional empathy (81 mentions), the recognition of nonverbal information (67 mentions), and the handling of complex cases (63 mentions). In addition, 63.6% of students explicitly indicated that they would recommend AI-assisted training to their peers. Conclusion: Generative AI was perceived by clinical medicine students as a useful supplementary resource for psychological counseling technique training, particularly for simulated interaction, repeated practice, and immediate feedback. However, students also recognized clear limitations concerning emotional empathy, nonverbal communication, and complex situations. These findings support a blended human–AI training model in which generative AI supplements standardized and repetitive practice while human instructors and supervisors retain central responsibility for professional judgment, relational learning, ethical guidance, and complex case supervision.