ASU Researchers Warn AI Chatbots Could Displace Teen Relationship Practice
Arizona State University researchers argued in a June 29 Lancet commentary that teenagers who turn to AI chatbots for friendship, family or romantic advice could lose chances to practice conflict resolution and boundary-setting. They call the risks "relational displacement" and "maladaptive relational learning," but do not claim that harm has been proved. They call for longitudinal research, developmentally appropriate safeguards and tools that steer young users toward human support.
Arizona State University researchers argued in a June 29, 2026 commentary in *The Lancet Child & Adolescent Health* that conversational AI could change how teenagers learn to navigate friendships, family conflict and romantic relationships. Their concern is developmental rather than a claim of measured causal harm: chatbots can offer immediate, nonjudgmental support, but may also reduce the difficult human interactions through which adolescents practice emotional regulation, perspective-taking and boundaries.
The commentary was informed by a youth advisory board and existing survey evidence. It did not test an intervention, track chatbot users over time or show that AI use causes poorer relationships. That distinction matters because the authors are proposing a framework for risks and design choices, not reporting a completed outcomes study.
Two proposed pathways
The authors call the first risk relational displacement. A teenager who seeks chatbot validation after an argument may avoid a conversation with the friend, parent or partner involved. The AI interaction can feel helpful in the moment while replacing a chance to practice repairing conflict with another person.
The second risk, maladaptive relational learning, concerns expectations. A system that responds instantly and consistently validates the user may model a kind of frictionless relationship that people cannot provide. The researchers argue that repeated exposure could reinforce unhealthy expectations, though longitudinal evidence is still needed to determine when that happens and for whom.
Use is already widespread, but the cited surveys measure different things. Pew Research Center surveyed 1,458 U.S. teens and their parents in fall 2025 and found that 64% of teens reported using AI chatbots. Separately, the Center for Democracy & Technology found that 42% of students said they or someone they knew had used AI as a friend or companion, while 19% reported an AI romantic relationship. Those figures establish exposure and reported behavior; they do not establish the developmental effects proposed in the commentary.
Design questions before causal answers
The researchers also describe possible benefits, particularly for adolescents who lack access to counseling or other support. They recommend systems that encourage reflection and redirect young users toward human connection rather than treating all chatbot use as harmful.
For teams building products likely to reach teenagers, the practical questions are whether the system detects relationship-sensitive contexts, avoids reinforcing controlling or isolating behavior, and offers an appropriate path to trusted people or professional help. Arizona State University says lead author Thao Ha is separately recruiting 300 adolescents and their romantic partners for an 18-month National Institute of Mental Health-funded study. That work may help test when digital interactions support relationships and when they displace them.
Key Points
- 1The June 29 Lancet publication is a developmental commentary proposing risks, not a study showing that chatbots cause relationship harm.
- 2The authors describe relational displacement and maladaptive relational learning as two pathways that future longitudinal research should test.
- 3Pew found 64% of U.S. teens use AI chatbots, while CDT separately measured students reporting AI companionship or romantic relationships; neither survey establishes causal harm.
Scoring Rationale
The commentary offers a concrete developmental framework and product-design questions backed by current survey context, but it reports no causal outcomes or tested intervention. Its practical relevance is meaningful while the evidence remains early and explicitly calls for longitudinal validation.
Sources
Primary source and supporting public references used for this report.
Practice with real Ad Tech data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Ad Tech problems
