Youth who used generative artificial intelligence (AI) for emotional support may have a higher prevalence of emotional issues.
Investigators analyzed data from the Ontario Health and Peer Relations Study collected from May 2025 through March 2026 across 4 school boards in Ontario, Canada. The analysis included 39,761 students in grades 4 through 12 with a mean age of 12.8 years, 49% of whom were female, 48% of whom were male, and 3% of whom were gender diverse. About 21% reported using generative artificial intelligence (AI) for emotional support or interpersonal advice, while 43% reported using AI for school-related tasks at least once.
Affective AI use involved utilizing an AI chatbot for help with feelings or advice about relationships with friends, parents, or other people, whereas functional AI use reflected utilizing AI for schoolwork. Emotional issues were assessed using 7 items from the Ontario Child Health Study Emotional Behavioral Scales, with a validated cutoff identifying clinically significant symptoms. Models accounted for demographic characteristics, perceived mattering, loneliness, functional AI use, and school-level clustering.
Clinically significant emotional issues were present in 58% of students who reported affective AI use vs. 29% of nonusers. After adjustment for demographic characteristics, loneliness, mattering, and functional AI use, affective AI use remained associated with a 26% higher prevalence of clinically significant emotional issues. Affective AI use also remained associated with higher continuous emotional-problem scores in the fully adjusted analysis.
Functional AI use showed a smaller dose-response association with emotional issues. Compared with students who never used AI for schoolwork, increases in emotional-problem scores ranged from 0.01 among those reporting a little functional use to 0.04 among those reporting a lot of use.
Affective AI use varied across demographic and developmental groups. It was reported by 27% of female students, 14% of male students, and 34% of gender-diverse students. Use increased with school level, from 16% among elementary students to 21% among middle school students and 26% among secondary school students. Students who were Black, Indigenous, or identified with other racial and ethnic groups also reported affective AI use more frequently than White students (24% vs. 18%).
The association between affective AI use and emotional issues varied by gender identity. The association was attenuated among gender-diverse students compared with female and male students, while the investigators found no difference between female and male students. School level did not modify the association, suggesting that the relationship between affective AI use and emotional issues was observed across the school levels studied.
The study's cross-sectional design precluded establishing temporal sequence or causality. The investigators could not determine whether youth with preexisting emotional distress were more likely to seek support from generative AI or whether reliance on AI for emotional support contributed to symptoms over time. Affective AI engagement was measured only as use vs.nonuse, preventing assessment of frequency, duration, or dose-response associations. Other limitations included reliance on self-reported measures, potential self-selection and residual confounding, and the absence of data on socioeconomic status and mental health treatment history. The findings involved general-purpose AI and may not be generalizable to specialized clinical AI agents.
The findings suggested that the motivation underlying AI engagement may matter more for youth mental health than technology use alone, with affective use showing a stronger association with emotional issues than functional use.
“Clinicians should normalize screening for affective AI use as this behavior serves as a robust marker for underlying mental health concerns,” wrote lead study author Kamilla Bonnesen, PhD, of the University of Ottawa, and colleagues.
Full disclosures of the study authors can be found in the study.
Source: JAMA Pediatrics
