Randomized Trial of AI Therapy Chatbot Shows 0.3 SD Mental Health Improvement Over 6 Months
A randomized controlled trial of the Mindsurf app—a CBT-based AI conversational agent—tested on Mexican women showed a 0.3 standard deviation improvement in mental health over six months, with no evidence of increased severe cases. Beyond mental health scores, treated participants also improved in sleep quality, healthful behaviors, daily functioning, and labor market outcomes (reduced work absences). Critically, treated participants were more likely to seek traditional psychotherapy, suggesting the AI intervention acts as a gateway rather than a replacement. Treatment effects persisted after the short engagement period ended.
Why It Matters
A peer-reviewed RCT showing persistent mental health improvements from an AI therapy chatbot—with positive spillover effects on employment and sleep—provides the strongest evidence yet for the clinical utility of CBT-grounded conversational AI at scale, particularly in underserved populations.