The study, led by corresponding author Tracy Vaillancourt of the Faculty of Education at the University of Ottawa, examined whether what the researchers describe as affective generative artificial intelligence use, meaning the use of AI tools for emotional rather than practical purposes, was associated with poorer mental health among youth. The central question was whether this behavior merely overlaps with known correlates of distress, such as loneliness or a diminished sense of mattering to others, or whether it represents something separate that clinicians and educators should pay attention to in its own right. The findings point to the latter: emotional reliance on AI emerged as a unique marker of psychological distress, independent of both mattering and loneliness.

This distinction matters because loneliness and a lack of mattering have long been recognized as powerful predictors of poor mental health outcomes in young people. Loneliness captures the painful gap between the social connection a person wants and the connection they actually experience, while mattering reflects the sense of being significant to others, of being noticed and valued by the people around them. Both constructs are closely tied to depression, anxiety, and suicidal ideation in adolescents. If turning to AI for emotional support were simply a symptom of loneliness or of feeling invisible, it would arguably add little new information. The fact that the association with distress persists even after accounting for these factors suggests that affective AI use is tapping into something additional about a young person’s inner life.

The researchers frame their conclusions around a practical and increasingly urgent distinction: the difference between functional AI assistance and the use of algorithmic interfaces as a digital refuge for emotional needs. Functional assistance covers the many legitimate and often beneficial ways young people interact with generative AI, such as getting help with homework, brainstorming ideas, answering factual questions, or learning new skills. Affective use, by contrast, involves bringing emotional burdens to a machine, asking a chatbot for comfort, confiding in it about painful feelings, or treating it as a substitute source of understanding and support. The study suggests that clinical frameworks and digital literacy programs should draw a clear line between these two patterns of behavior, because they may carry very different implications for a child’s wellbeing.

The timing of this research is significant. Generative AI tools have moved from research laboratories into everyday life at remarkable speed, and children and adolescents are among their most enthusiastic adopters. Conversational agents are now available around the clock, respond instantly, never judge, and can simulate empathy with convincing fluency. For a young person who feels isolated, misunderstood, or burdened, the appeal of such an interlocutor is easy to understand. Yet the same qualities that make these systems attractive also raise questions about what it means when a developing mind begins to route its emotional needs through an algorithm rather than through family, friends, teachers, or mental health professionals.

From a clinical standpoint, the study’s findings carry implications for how practitioners assess and respond to psychological distress in youth. If asking an AI for emotional support is a unique marker of distress, then simply asking a young patient whether they use AI tools, and how they use them, could provide clinicians with diagnostically meaningful information that might otherwise go uncollected. A teenager who reports using chatbots primarily for schoolwork presents a very different picture from one who reports turning to them when feeling sad, anxious, or alone. The authors argue that clinical frameworks should be updated to make this distinction explicit, so that affective AI use is recognized and explored rather than overlooked or lumped together with benign technology use.

The implications extend beyond the clinic into schools and public health more broadly. Digital literacy programs have traditionally focused on topics such as online safety, privacy, misinformation, and screen time. The present findings suggest that these programs should also help young people understand the difference between using AI as a tool and using it as an emotional refuge. Teaching children and adolescents to recognize when they are seeking comfort from a machine, and to reflect on what that impulse might be telling them about their own emotional state, could turn a passive risk factor into an opportunity for self-awareness. Equally, such programs could help adults, including parents and teachers, understand that a child’s growing attachment to conversational AI may be less about fascination with technology and more about unmet emotional needs.

The study is published alongside an editorial and an editor’s note in the JAMA Network, reflecting the significance that the journal’s editors attach to the topic. Cross-sectional research of this kind captures a snapshot in time, which means it can identify associations but cannot establish whether emotional distress drives young people toward AI, whether heavy affective AI use contributes to distress, or whether both patterns are shaped by other underlying factors. The authors are careful to describe the relationship as a marker rather than a cause, a distinction that matters for interpretation. What the design does establish is that the association is real and independent: even when loneliness and mattering are held constant, young people who seek emotional support from generative AI show higher levels of psychological distress.

That independence is the study’s most striking contribution. It suggests that affective AI use is not reducible to the social deficits that mental health professionals already screen for. A young person might report feeling connected to others and valued by them, yet still turn to an algorithm for emotional support, and that behavior alone would flag elevated distress. This makes the behavior potentially valuable as an early signal, something visible on the surface of everyday technology use that hints at struggles happening beneath it. In an era when parents and educators often struggle to detect internalizing problems in children, precisely because conditions such as depression and anxiety tend to be hidden rather than disruptive, a marker embedded in routine digital behavior could prove genuinely useful.

The broader conversation about AI and youth mental health is only beginning, and this study adds an important empirical data point to a debate that has so far been dominated by speculation. As generative AI becomes more emotionally sophisticated, more personalized, and more deeply woven into the daily routines of young people, the questions raised by this research will only grow in urgency. The authors’ central message is measured rather than alarmist: the goal is not to condemn AI use among youth, much of which is functional and harmless, but to recognize that when children and adolescents begin treating algorithmic interfaces as a refuge for their emotional lives, adults should pay attention. Distinguishing between a young person using AI to finish an assignment and one using it to fill an emotional void may become one of the more consequential skills for the clinicians, educators, and families raising children in the age of artificial intelligence.

Subject of Research: The association between affective use of generative artificial intelligence and psychological distress in children and adolescents

Article Title: Affective generative artificial intelligence use and youth mental health

Article References: Affective generative artificial intelligence use and youth mental health. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: generative AI, youth mental health, psychological distress, loneliness, mattering, JAMA Pediatrics, adolescents, digital literacy, chatbots, emotional support, cross-sectional study, clinical frameworks