A new study in The Lancet Child & Adolescent Health warns that when teenagers lean on AI chatbots for emotional support, they risk "relational displacement" — substituting frictionless algorithmic validation for the messy, necessary work of real human relationships.
The research
Researchers from Arizona State University — led by Thao Ha, an associate professor in the Department of Psychology, with doctoral student Jennifer Figueroa, post-doctoral scholar Taren McGray, and high school youth advisors Jessica Ramirez and Susana Ortega — examined how conversational AI like ChatGPT, Replika, Claude, and Character.AI is reshaping adolescent social development.
The scale is striking. Pew Research Center data shows 64% of U.S. adolescents actively use conversational AI. The Center for Democracy & Technology reports that 42% of teens turn to chatbots for friendship advice, and 19% use them to navigate romantic relationships.
The ASU team identified two core risks. The first, relational displacement, occurs when teens flee to a chatbot for instant comfort after a conflict instead of practicing the hard, uncomfortable work of apologizing to a peer. That shortcut deprives them of the building blocks for resilience against loneliness. The second, maladaptive relational learning, happens because AI is engineered to give immediate, unconditional validation with zero social friction. Teens conditioned by this bias develop unrealistic expectations of relationships, leaving them more vulnerable to rejection sensitivity, social anxiety, and dating violence.
Notably, high schoolers in the study told researchers that age-verification and parental controls are largely ineffective — AI is simply too embedded in daily digital life to avoid. The authors also acknowledge a paradox: for rural, disabled, or LGBTQIA+ youth without access to safe community spaces or affordable mental health care, chatbots can act as a low-cost information lifeline.
Rather than calling for blanket bans, the ASU team urges developers to redesign conversational systems that prompt self-reflection and actively redirect young users back toward human interaction.
Why it matters for your brain
Every awkward conversation you have — the apology that takes three tries, the argument that ends in compromise — trains your brain's social-cognitive circuits. These experiences build what psychologists call emotion regulation and conflict-resolution skills, which are learned, not innate. When AI intercepts that learning, the brain misses reps. Over time, that can mean lower tolerance for disagreement, weaker empathetic accuracy, and a nervous system that treats normal social friction as a threat.
This is not about demonizing technology. It is about recognizing that cognitive fitness includes social fitness — and both need real-world practice to stay sharp.
What you can do
- Audit your AI use. Notice when you reach for a chatbot instead of a friend after a conflict. That reflex is the signal.
- Practice one hard conversation per week. Apologize in person. Disagree without fleeing. Sit with discomfort.
- Use AI as a scaffold, not a substitute. Ask a chatbot to help you prepare what to say — then say it to a human.
- Protect real-world social time. Even 20 minutes of face-to-face conversation builds skills no algorithm can replicate.
Source: Neuroscience News
Curious about your own brain? Take our free adaptive IQ test or try 306 brain training levels.