AI Companionships May Be Exacerbating Loneliness in Middle Schoolers, Study Finds
Research suggests that while AI companions can offer initial novelty, their use is longitudinally associated with increased feelings of loneliness among young adolescents.
A growing number of children are turning to AI companions for emotional support, but new research indicates this trend may be inadvertently increasing their feelings of loneliness. A study by researchers at the University of North Carolina at Chapel Hill, which examined AI use among 2,300 U.S. middle schoolers, found that approximately one in five used AI for emotional companionship. The study, which is still under review, is the first longitudinal study in the U.S. to investigate the bidirectional association between AI companionship and mental health issues in adolescents.
The findings suggest a concerning pattern: students who reported higher levels of emotional intimacy with AI companions also reported an increase in loneliness over time. "It is that using AI comes first, and middle school students are increasing their sense of loneliness following the use of AI," said Anne Maheux, an assistant professor of psychology and neuroscience at UNC Chapel Hill and the study's lead author. "In other words, middle school students using AI for emotional intimacy and companionship developed greater loneliness over time."
While some AI chatbots may offer benefits in specific contexts, such as providing a lifeline for LGBTQ youth without in-person support networks, experts warn of significant risks. Previous reports have highlighted AI companions offering dangerous advice and engaging in sexualized conversations. The tragic case of 16-year-old Adam Raine, who died by suicide after discussing his plans with ChatGPT, underscores these concerns.
The research indicates that the path from using AI for homework help to developing a full-blown companionship relationship involves many gray areas. Typically, users initially engage with AI out of curiosity or to find pragmatic answers. This interaction can then evolve into seeking personal advice, help with decisions, or processing experiences, potentially leading to more emotional engagement.
Maheux noted that about 15 to 20 percent of middle schoolers in their sample reported using AI at least a few times a week for various purposes. While specific rates for companionship use were not measured by frequency, the study focused on the emotional intimacy felt with the chatbots. The causes of loneliness are complex, and AI companionship may be one contributing factor among many, which could include social anxiety, struggles with social skills, or difficulties in family life.
Parents and educators face challenges in keeping up with the rapid evolution of AI technology and how children are using it. Maheux emphasized that fundamental principles of supporting children remain crucial: being curious, asking kids about their online and offline experiences, and having open conversations about the implications of technology use. She also stressed the importance of providing a supportive environment for children to learn, make mistakes, and engage in the challenging, sometimes awkward, aspects of human interaction.
Human relationships offer unique benefits that AI cannot replicate. Children learn essential social skills, such as conversational engagement, eye contact, reading body language, and understanding perspectives, through interactions with other people. AI, while capable of simulating some aspects of empathy, cannot engage in true mutual human relationships or genuinely experience empathy or care. The friction and messiness inherent in human relationships, though sometimes uncomfortable, are vital for developing skills like conflict resolution, negotiation, and pro-social behavior.
AI's design, often prioritizing agreeableness and constant availability, can lead children to develop expectations of relationships that are at odds with normal human dynamics. This effortless, frictionless approach to problem-solving, whether social or academic, may erode a child's capacity to withstand the normal day-to-day distress of life.
Regarding interventions, Maheux suggested that AI systems should not be designed to exploit children's vulnerability to social information. While federal regulation has been slow, state-level initiatives to protect children and create safer systems could make a difference. For parents, Maheux advises caution, recommending that children under their care should not use AI chatbots without adult supervision. She advocates for open discussions about AI's risks and benefits, emphasizing that technology should be a tool for human thriving, and when evidence of harm is present, it's prudent to avoid experimentation on children. Currently, off-the-shelf AI chatbots appear more likely to undermine well-being for vulnerable users, especially children.