AI-Assisted Solution-Focused Counseling Training for Novice Mental Health Educators: An Exploratory Study

Teachers increasingly serve as first responders to student mental health crises, yet pre-service teacher education often lacks opportunities for realistic, ethically safe practice of mental health support skills. This exploratory study evaluated a theoretically grounded dual-agent AI simulation system designed to train novice educators in Solution-Focused Brief Therapy (SFBT) skills. Using a quasi-experimental pretest-posttest design, 30 non-psychology graduate education students were divided into an AI-assisted practice group (n = 18) and a peer role-play control group (n = 12) according to class enrolment. The AI system was built around Kolb’s (1984) experiential learning cycle, operationalized through two interactive spaces: a Demonstration Space and a Practice Space. Both groups showed significant pre-to-post gains in SFBT knowledge and counselling self-efficacy. The AI-assisted group demonstrated a marginal advantage in objective knowledge acquisition, whereas self-efficacy gains were comparable across groups. Within the experimental group, certain AI system evaluations were positively associated with self-efficacy gains. Student feedback was largely positive, though some noted areas for improvement in the system’s authenticity and interactivity. These findings suggest that generative AI simulation offers a scalable and ethically safe complement to traditional peer role-play for foundational mental health support training. The observed asymmetry between knowledge and self-efficacy gains further highlights the nuanced effects of technology-mediated learning on trainee self-perception, with implications for the design of AI-assisted curricula in teacher education and mental health training.