Clinicians’ Reflections on Explainable AI (XAI) in Counselor Education and Diagnostic Training: An Exploratory Qualitative Study

Current approaches to teaching clinical diagnosis in counselor education and supervision often rely on self-reports, brief vignettes, and supervisors’ recollections—methods that can be resource- intensive and prone to bias, contributing to diagnostic inaccuracy among trainees. While artificial intelligence (AI) has gained traction in healthcare, many machine learning (ML) models function as “black boxes,” limiting transparency and educator trust. Explainable AI (XAI), which focuses on making AI processes interpretable, offers a potential solution for counselor training. This exploratory qualitative study examined eight licensed psychologists’ reflections after viewing a live demonstration of an original XAI model trained to predict depression. Using Reflexive Thematic Analysis, we identified five themes describing supervisors’ perceptions of XAI’s potential to enhance diagnostic transparency, facilitate reflective supervision, and support trainee learning—alongside concerns about data quality, ethics, and the preservation of human judgment. The findings offer preliminary insights into how XAI might be responsibly integrated into counselor education and supervision to complement, rather than replace, the relational and ethical dimensions of training.