The impact of artificial intelligence-based clinical training on burnout among practitioners treating serious mental illness.

Burnout is a pervasive challenge in the mental health field, particularly among practitioners treating serious mental illness. Although skills-based training may reduce burnout, many programs are not equipped to support continued engagement. Recent advances suggest artificial intelligence (AI) may address these limitations. This study compared provider burnout after receiving standard web-based didactic training (treatment as usual [TAU]) to TAU plus an AI-supported training platform (TAU + AI) that included immediate feedback. Participants were frontline behavioral health providers working with clients with serious mental illness, where burnout was assessed at baseline and 3-month follow-up. Across groups, burnout significantly increased from pre- to posttraining, t(95) = 3.72, p < .001, Cohen's d = 0.38. There were no significant differences in burnout change between the TAU + AI and TAU conditions, suggesting brief AI-supported interventions may be insufficient to alleviate burnout. Secondary analyses revealed no association between frequency of practice and burnout change. However, there was a significant negative relationship between posttraining burnout and the proportion of practices that met fidelity benchmarks. A significant interaction effect also indicated that meeting fidelity benchmarks predicted lower burnout only when those exercises were completed at a high level of proficiency. Together, these findings suggest that training may impact burnout only if learners successfully acquire new skills, potentially improving confidence and self-efficacy. While training alone may not be sufficient to reduce burnout, integrating AI-based feedback and skill assessment into training may allow systems to better verify and support provider skill acquisition and ultimately reduce burnout. (PsycInfo Database Record (c) 2026 APA, all rights reserved).