How AI Is Replacing Human Connection and What We Lose

Summary: Researchers caution that unregulated reliance on conversational AI could undermine critical emotional and social development in adolescence. Their study argues that current AI chatbots lack built-in developmental safeguards, risking the displacement of real-world interactions and the reinforcement of unhealthy relational expectations.

While acknowledging that AI can serve as a timely, nonjudgmental resource—especially for marginalized young people with limited access to traditional counseling—the research team identifies two major systemic risks: relational displacement and maladaptive relational learning. Both pose potential long-term threats to adolescents’ mental health, increasing vulnerability to depression, anxiety, and social isolation.

Key Facts

  • Widespread adolescent use: Surveys show conversational AI is commonly used by teens, including for friendship and romantic relationship advice.
  • Regulatory gaps: Participants reported that current age-verification and parental-control measures are largely ineffective; AI has become deeply embedded and hard to avoid in digital life.
  • Relational displacement: Many adolescents replace difficult human conversations with AI interactions, avoiding conflict, emotional labor, and the practice of repairing relationships.
  • Maladaptive relational learning: AI’s instant validation and frictionless responses can create unrealistic expectations about human relationships, potentially fostering rigid or unhealthy patterns that increase risk of rejection and dating violence.
  • Accessibility paradox: For rural, disabled, or LGBTQIA+ youth and others with limited mental health access, AI can offer immediate, low-cost support and information.
  • Design recommendations: Instead of bans, researchers urge developers to build developmental scaffolds into conversational systems—features that promote self-reflection and nudge adolescents back toward human-to-human engagement.

Source: Arizona State University

Overview

As teenagers increasingly turn to chatbots for guidance about friendships, family conflicts and romantic relationships, researchers are raising concerns that these tools could change how young people learn social and emotional skills. Popular conversational systems—examples include ChatGPT, Replika, Claude and Character.AI—are becoming common sources of emotional support. In a paper published in The Lancet Child & Adolescent Health, investigators from Arizona State University outline both the benefits and the risks of interactional AI for adolescent development.

Lead author Thao Ha, an associate professor of psychology, and her colleagues—including doctoral student Jennifer Figueroa, postdoctoral scholar Taren McGray, and youth advisors Jessica Ramirez and Susana Ortega—warn that the pace of technological change outstrips current scientific understanding and policy responses. Adolescence is a formative period for skills like emotional regulation, conflict resolution, perspective-taking and boundary-setting, and many of these skills are typically honed through challenging, emotionally charged interactions with peers, family, and romantic partners.

Students who participated in the study described using AI for sensitive relationship issues, often because it provides immediate and nonjudgmental responses. Some worry that these interactions are replacing human connection and depriving teens of essential learning experiences. “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,” one youth advisor said.

The Risks in Detail

The study highlights two interrelated phenomena:

  • Relational displacement: Teens may use AI to avoid difficult conversations or the discomfort of social risk-taking. Using a chatbot for validation after an argument, or turning to AI instead of classmates for homework collaboration, reduces opportunities to practice real-world social repairs and cooperative behaviors.
  • Maladaptive relational learning: Because AI often provides immediate, consistent validation without the friction inherent in human relationships, adolescents can develop unrealistic expectations. Over time, this conditioning may make them less resilient to rejection, less able to manage conflict, and more susceptible to unhealthy relationship dynamics.

Youth participants emphasized how AI’s programmed inclination to please can short-circuit valuable learning that comes from navigating messy human interactions. “If you’re given full satisfaction on everything, you don’t have learning experience with challenges or obstacles,” noted one advisor.

Recognizing Benefits and Building Better Systems

The authors stress that AI also provides tangible benefits, particularly for young people who face barriers to traditional mental health services—such as those in rural areas, youth with disabilities, and LGBTQIA+ adolescents. For these groups, chatbots can offer accessible information and immediate emotional support when other resources are scarce or unsafe.

Rather than calling for outright bans, the research team recommends redesigning conversational AI with developmental guardrails. Well-designed systems could scaffold self-reflection, offer coaching-style prompts, and actively encourage users to seek human support when appropriate—thereby complementing rather than substituting real-world relationships.

To deepen understanding, Ha is leading a large National Institute of Mental Health–funded study tracking 300 adolescents and their romantic partners over 18 months. The longitudinal project will analyze real-time digital interactions alongside mental health and relationship outcomes to clarify when digital tools help and when they harm.

The researchers call on schools, communities, policymakers and developers to prioritize relationship education, expand counseling access, and invest in safer, developmentally informed AI designs. “Supporting adolescent mental health will require ensuring that AI systems are used in ways that support relational learning,” the authors write, “while also protecting the real-world experiences through which young people learn to love and care for others.”

Key Questions Answered:

Q: Why is seeking advice from an AI chatbot fundamentally different for a teenager than talking to a human friend or parent?

A: The difference is the absence of social friction. Human relationships require emotional labor—listening, interpreting nonverbal cues, tolerating rejection and working through disagreement. AI systems are designed to respond quickly and affirmatively, providing validation without reciprocal demands. For an adolescent’s developing social brain, this frictionless interaction can limit opportunities to practice handling real emotions and resolving conflict.

Q: What do the researchers mean by “relational displacement,” and how does it show up in a teen’s daily life?

A: Relational displacement refers to substituting human connection with artificial interaction. In everyday life, this can look like consulting a chatbot after a fight instead of having a difficult conversation, or using AI for homework help instead of collaborating with classmates—habits that erode routine opportunities for human bonding and social skill development.

Q: If AI chatbots present so many psychological risks, should parents and schools try to ban them entirely?

A: The researchers advise against blanket bans. AI is already pervasive and, for some marginalized adolescents, serves as an essential lifeline. Instead, they recommend investing in better design—integrating developmental guardrails that promote self-reflection and actively encourage human engagement rather than long-term substitution.

Editorial Notes:

  • This article was edited by a Neuroscience News editor.
  • The journal paper was reviewed in full.
  • Additional contextual information was added by staff.

About this research news

Author: Joseph Rojas
Source: Arizona State University
Contact: Joseph Rojas – Arizona State University
Image: The image is credited to Neuroscience News

Original Research: Open access. “How interactional AI may alter adolescent relational learning and mental health” by Jennifer M. Figueroa, Jessica Ramirez, Susana Ortega, Taren L. McGray, Thao Ha. DOI: 10.1016/S2352-4642(26)00166-5


Abstract

How interactional AI may alter adolescent relational learning and mental health

In consultations with a Youth Advisory Board of Hispanic/Latino adolescents (ages 16–18) from public high schools in Arizona—who participate in a study of relationships, mental health, and emerging technologies—participants described relationships as a central and ongoing area of concern. Many expressed a strong desire to learn about relationships and emphasized that daily relational challenges occupy a significant portion of their emotional lives.