A systematic review and three-level meta-analysis evaluated the effects of AI-based chatbot interventions on mental health outcomes (depression, anxiety, loneliness, affect, sleep) across 16 eligible studies yielding 67 effect sizes.
AI-based chatbot interventions were associated with a small-to-moderate improvement in mental health outcomes (Hedges' g = 0.47, 95% CI [0.18, 0.77]); significant improvements were observed across depression, anxiety, loneliness, affect, and sleep, with a stable pooled estimate on leave-one-out sensitivity analysis.
Most included studies presented some risk of bias; overall evidence certainty was rated low; no significant moderators were identified, limiting guidance on which chatbot types or populations benefit most.
AI-based chatbots show modest but statistically significant benefits across several mental health domains and may serve as scalable supplementary tools where access to care is limited. Clinicians should interpret results cautiously given low certainty of evidence and treat chatbots as packaged interventions rather than standalone AI-specific therapies.
Explore related topics
