THERAPEUTIC RELATIONSHIP WITH MACHINES: A REVIEW OF CLINICAL TRIALS AND PSYCHOSOCIAL OUTCOMES OF AI-POWERED CHATBOTS IN MANAGING DEPRESSION AND ANXIETY
TL;DR
Meta-analysis results demonstrated that AIpowered chatbots were effective in reducing social anxiety, and therapeutic alliance with machines reached a mean of 3.84 within five days, comparable to human-led therapies.
Abstract
Mood and anxiety disorders impose a substantial burden on public health, and limited access to mental health professionals has intensified the need for digital tools [1]. These technologies contribute to improved psychological outcomes through cognitive restructuring and facilitation of emotion regulation processes [2]. The present study aims to investigate the effectiveness and psychosocial outcomes of AIpowered chatbots in managing depression and anxiety. Inclusion criteria comprised clinical trials and interventional studies involving adolescents and adults with clinical symptoms, while exclusion criteria included technical studies and reports lacking clinical data. A systematic search was conducted in the OpenAlex database covering the period from 2020 to 2025. The selection process began with 955 records, and following literature screening, 28 studies were selected for data extraction. Data extraction focused on primary outcomes including depression and anxiety scores, as well as therapeutic alliance indicators. Risk of bias assessment was performed using validated tools, and findings were synthesized through narrative synthesis and meta-analysis methods. Meta-analysis results demonstrated that these interventions were effective in reducing social anxiety, with a Hedges' g effect size of 0.36 and a 95% confidence interval ranging from 0.29 to 0.43, with 0% heterogeneity [3]. Furthermore, a reduction in depressive symptoms yielded an effect size of 0.64 with a 95% confidence interval between 0.42 and 0.66 and heterogeneity of 95% [4]. Findings elucidated that therapeutic alliance with machines reached a mean of 3.84 within five days, comparable to human-led therapies [5]. The paucity of valid longitudinal data and reliance on selfreported user assessments constitute the primary limitations of the existing evidence in this domain. AIpowered chatbots are recognized as effective complements to traditional therapies and facilitate equitable access to mental health care across diverse populations.