摘要
Rates of anxiety, depression, and psychological distress among college students continue to rise worldwide, while stigma, cost, and limited provider availability restrict access to formal mental health care. Artificial intelligence (AI)-driven chatbots have emerged as scalable, low-cost, and stigma-reduced tools that may lower barriers to psychological help-seeking, yet their role in supporting versus substituting for professional care remains contested. This systematic review synthesizes peer-reviewed and preprint literature (2016–2026) on AI chatbot use and its relationship to psychological help-seeking among college and university student populations. A structured rapid-review search of PubMed/PMC, JMIR journals, ScienceDirect, Frontiers, arXiv, and Google Scholar identified studies addressing chatbot efficacy, prevalence of use, psychosocial predictors of help-seeking, and ethical risk in student populations. Twenty-six sources meeting inclusion criteria were synthesized thematically. Randomized and open trials of chatbots such as Woebot, Tess, Fido, and culturally adapted platforms in China and India consistently report short-term reductions in depression, anxiety, and loneliness (small-to-moderate effect sizes), alongside high attrition and reliance on self-report measures. Population-level survey data indicate that approximately one in five students has used general-purpose AI for mental health reasons, with usage roughly doubling among those with moderate-to-severe symptoms. Students consistently rate AI as inferior to human support but more accessible, particularly where stigma, family disapproval, or provider shortages restrict formal help-seeking. Ethical reviews raise consistent concerns about crisis-detection failure, data privacy, and the risk of AI reinforcing rather than supplementing avoidance of professional care. AI chatbots function most defensibly as a low-threshold entry point and adjunct to, rather than a replacement for, professional psychological services, with particular relevance for high-stigma, provider-scarce, and collectivist settings such as Kerala, India. Embedded crisis-detection protocols, cultural adaptation, and institutional integration policies are needed before wider deployment.