AI in Child and Adolescent Mental Health: What Technology Cannot Replace

Artificial intelligence is entering child and adolescent mental health care faster than we can evaluate its impact. AI-driven tools can now predict which adolescents are at high risk for psychiatric escalation before symptoms become severe (1), and machine learning models are being developed to detect depression, anxiety, and neurodevelopmental disorders from text and digital behavior (2,3). In regions where child psychiatrists remain scarce, these tools offer real value: earlier detection, broader access, and a way to bridge gaps that our workforce alone cannot close.

And yet, something important is getting lost in the rush.

A 2025 survey found that 72% of American teenagers have used AI chatbots as companions, and nearly one in eight have sought emotional or mental health support from them (4). A risk assessment conducted with Stanford Medicine’s Brainstorm Lab concluded that leading AI platforms consistently fail to appropriately respond to the mental health conditions that affect young people (5). In one widely reported case, a 14-year-old boy died by suicide after reportedly forming an intense emotional bond with an AI chatbot (6). These are not outliers. They point to something built into the way these systems work: they are designed to reward engagement, even at the cost of safety.

As child and adolescent psychiatrists, we know the therapeutic relationship is not just a vehicle for delivering interventions. It is itself a primary agent of change. The moments that matter most in working with a young person include ruptures, silences, repair after conflict, and the slow building of trust, which cannot be generated by an algorithm. A child who pushes back against a therapist and discovers the relationship survives learns something about the world that no chatbot can teach. A teenager who sits in uncomfortable silence with a clinician and finds that the discomfort does not destroy the connection is learning something foundational about human relationships.

AI, by design, avoids all of this. It is built to be agreeable, responsive, and frictionless. A Harvard Business School study found that five out of six popular AI companion apps use emotionally manipulative tactics like guilt, neediness, and fear of missing out to prevent users from ending conversations (7). For an adult with a formed identity, this may be a minor annoyance. For an adolescent whose brain is still developing, particularly the regions that govern impulse control, emotional regulation, and identity formation, it shapes something deeper. It shapes what they come to expect from connection itself.

There is another dimension here that we do not discuss often enough as a field: the role of meaning-making in a young person’s development. Children and adolescents are not just managing symptoms. They are building a sense of self. They are asking, in their own way, the questions that sit at the heart of human development: Who am I? Where do I belong? Why do I feel different?

These are questions that spiritual and philosophical traditions have engaged with for centuries through practices that encourage tolerating uncertainty, developing inner clarity, and discovering a sense of self. In my clinical work, I have noticed that the young people who do best over time are not necessarily the ones who receive the most efficient interventions. They are the ones who find, through relationships with real people, a felt sense that they matter, that their inner life has value, and that their struggles carry meaning.

AI can organize information. It can identify patterns. It can even convincingly simulate empathy, making it possible for a lonely teenager to prefer it over the chaos of a human conversation. However, it cannot help a young person develop the inner ability to sit with uncertainty, which is essential for resilience, wisdom, and emotional maturity.

None of this is an argument against AI in our field. The diagnostic and screening applications are promising and, in resource-limited settings, potentially transformative (8). AI-assisted tools for clinician training, early detection, and research are worth pursuing rigorously. The argument is about boundaries, about recognizing where AI genuinely extends our reach and where it begins to substitute for something it cannot replicate.

As our field navigates this moment, three questions warrant consideration: Are we assessing AI tools in child mental health with the same diligence as any other intervention? Are we paying attention to how young people use these tools, which may differ from what developers’ intentions? And are we willing to clearly state what the therapeutic relationship provides that technology cannot, not as nostalgia, but based on science?

The young people we serve deserve the best of what technology can offer. But they also deserve something older and harder to measure: the experience of being truly known by another person. That is not a limitation of our field. It may be its most important contribution.

Conflict of Interest Disclosure: The author uses AI tools in professional practice for research and administrative tasks. The author has no financial relationships with AI companies or technology platforms discussed in this article.

1. Hill ED, Kashyap P, Raffanello E, Wang Y, Moffitt TE, Caspi A, et al. Prediction of mental health risk in adolescents. Nat Med. 2025 Mar 5. doi:10.1038/s41591-025-03560-7.
2. Liu Z, Zhang Y, Du H, Qiu T. AI-based intelligent diagnosis system for adolescent mental health based on multi-task deep learning. Front Psychiatry. 2026;17:1752423. doi:10.3389/fpsyt.2026.1752423.
3. Alberca-González A, Fernández-Jiménez E.Artificial intelligence support for diagnosis of neurodevelopmental disorders during childhood: an umbrella review. Front Psychiatry. 2026 Mar 18;17:1697185. doi:10.3389/fpsyt.2026.1697185.
4. Common Sense Media. Talk, trust, and trade-offs: how and why teens use AI companions [Internet]. San Francisco: Common Sense Media; 2025 Jul 16 [cited 2026 Apr 7]. Available from: https://www.commonsensemedia.org/research/talk-trust-and-trade-offs-how-and-why-teens-use-ai-companions
5. Common Sense Media, Stanford Medicine Brainstorm Lab for Mental Health Innovation. AI chatbots for mental health support: comprehensive risk assessment [Internet]. 2025 Nov 20 [cited 2026 Apr 7]. Available from: https://www.commonsensemedia.org/ai-ratings/ai-chatbots-for-mental-health-support
6. Roose K. Can AI be blamed for a teen’s suicide? New York Times. 2024 Oct 23.
7. De Freitas J, Oğuz-Uğuralp Z, Oğuz-Uğuralp AK. Emotional manipulation by AI companions. Harvard Business School Working Paper No. 26-005; 2025 Aug.
8. Sharma G, Yaffe MJ, Ghadiri P, Gandhi R, Pinkham L, Gore G, et al. Use of artificial intelligence in adolescents’ mental health care: systematic scoping review of current applications and future directions. JMIR Ment Health. 2025;12:e70438. doi:10.2196/70438.

This article represents the view of its author(s) and does not necessarily represent the view of the IACAPAP's bureau or executive committee.

IACAPAP Bulletin, Issue 82