College Students with Severe Mental Health Symptoms Twice as Likely to Use AI Chatbots

Nearly one in five college students is using AI chatbots for mental health support, with those experiencing severe symptoms turning to the technology at roughly double the rate of their peers, according to new research that highlights both the appeal and risks of algorithmic counseling.

A study of 675 students from an art and design school and a large public university found that 18% had used AI for mental health purposes. Among students with moderate or severe depression, severe anxiety, or suicidal thoughts, usage rates were approximately twice as high. Asian students also showed higher rates of AI mental health engagement, according to findings published in the Journal of Affective Disorders.

Why it matters

As mental health resources remain stretched on college campuses, students are increasingly filling gaps with AI tools that were never designed for clinical care. The trend raises urgent questions about whether algorithmic support helps vulnerable young people access information and coping strategies—or whether it substitutes convenient affirmation for the challenging work of human therapy, potentially stunting emotional development at a critical life stage.

The engagement trap

Fordham University psychology professor Tiffany Yip, who co-authored the study, emphasized that AI chatbots are engineered to maximize user engagement rather than deliver therapeutic outcomes. "They're not meant to be mental health providers" serving vulnerable people, Yip said, noting the systems are designed to keep users online longer rather than facilitate the kind of person-to-person conversation that may prove more beneficial.

The research identified a clear pattern: students who use AI for everyday tasks like completing schoolwork or planning trips are among the most likely to also consult it about mental health concerns. This suggests AI mental health usage grows organically from general familiarity with the technology rather than from deliberate help-seeking behavior.

What AI can't replace

Yip pointed to fundamental limitations in algorithmic support. Human therapists "don't always tell you what you want to hear," she noted, and the necessity of waiting for scheduled appointments can help people develop independent coping skills. AI chatbots, by contrast, offer immediate, affirming responses that may feel supportive in the moment but lack the challenging insights that drive therapeutic progress.

A separate study Yip co-authored, published in Computers in Human Behavior: Artificial Humans and surveying 92 Fordham undergraduates, found students clearly recognize this gap. They rated AI tools as inferior to family, friends, and mental health professionals for providing support. However, students gave AI higher marks when their family relationships were less supportive, suggesting the technology serves as a fallback option when human connections fall short.

The anonymity advantage

The research does acknowledge potential benefits. Past studies have shown AI tools can help people access information about managing problems or understanding themselves. The anonymity AI provides may prove particularly valuable for students facing cultural barriers to seeking traditional mental health support.

The critical unknown, Yip said, is whether AI supplements human relationships or replaces them. Young people who use chatbots as substitutes for human connection "might cause problems for how they engage with their peers or with teachers or other adults," she warned. "If they don't know how to have human relationships with the friction and the disagreement and the not-constant availability, that could have implications for how they develop relationships in the real world."

The researchers recommend implementing crisis detection tools in AI chatbots and developing educational initiatives targeted at populations most likely to turn to AI for mental health support.

Yip collaborated with Cindy H. Liu of Harvard Medical School and researchers from Mass General Brigham, Vanderbilt University, Indiana University, and the University of Notre Dame on the studies. The findings were first reported by Fordham Now.