Researchers Warn AI Use Could Impede Teen Relationships

Writing in *The Lancet Child & Adolescent Health*, Thao Ha, an associate professor of psychology at Arizona State University, and co-authors (doctoral student Jennifer Figueroa, postdoctoral scholar Taren McGray, and two teenage youth advisory board members) argue that AI-powered tools including ChatGPT, Replika, Claude and Character.AI are becoming a common source of emotional support for teenagers seeking advice on friendships, family conflict and romantic relationships. The authors say these tools can offer immediate, nonjudgmental guidance with real developmental benefits, but warn that without safeguards, reliance on them may bypass the person-to-person interactions through which teens typically learn emotional regulation, conflict resolution, perspective-taking and boundary-setting. "The technologies are developing super-fast, faster than we can keep up with as scientists, faster than governance and policy can keep up with," Ha said. The paper cites Pew Research Center survey data finding 64% of U.S. teens use conversational AI, and Center for Democracy & Technology data showing 42% have used chatbots for friendship-related purposes and 19% for romantic relationships.

The authors describe two specific risks rather than a single blanket harm claim. "Relational displacement" refers to teens substituting AI conversations for difficult but developmentally useful conversations with peers, family or partners - for example, seeking chatbot validation after an argument instead of resolving it with the other person. "Maladaptive relational learning" refers to AI's constant, frictionless validation potentially setting unrealistic expectations that human relationships then fail to meet, which the authors say could increase vulnerability to rejection, dating violence and mental health problems over time. Youth advisory board member Susana Ortega, a high school senior involved in the research, said: "We all mostly had concerns about how AI was replacing actual human connection and how it limits a lot of those needs that humans have that cannot be replaced with a computer artificial intelligence." The authors also note AI's potential benefits for teens who face barriers to traditional support, such as those in rural areas or without access to counseling, and say the goal is better-designed AI rather than discouraging use altogether.

This is a call to action rather than a set of proven metrics: the authors do not publish a tested intervention or engagement benchmark. Still, teams building or evaluating conversational agents likely to be used by teens should consider tracking whether interactions substitute for or supplement human relationships, building in prompts that redirect users toward offline support and human connection, and offering transparency or oversight mechanisms for parents, schools or clinicians. Ha is now leading a National Institute of Mental Health-funded study following 300 adolescents and their romantic partners for 18 months to examine when digital interactions help or harm relationships and mental health - a signal that firmer, longitudinal evidence is still years away.

Whether the NIMH-funded longitudinal study and other peer-reviewed research quantify developmental effects over time, and whether platforms or regulators move toward age-targeted design guardrails given that current approaches like age verification were described by teens in the study as ineffective.