This cross-sectional study examined whether using generative AI for emotional support (affective AI use) vs. school tasks (functional AI use) was associated with emotional problems in 39,761 students in grades 4–12 across Ontario, Canada (May 2025–March 2026).
21.1% of students used AI for emotional support. Affective AI use was associated with nearly double the rate of clinical emotional problems (57.7% vs. 29.2%; prevalence ratio, 1.98). After adjusting for loneliness, mattering, and demographics, the association persisted (B = 0.14), while functional AI use had only a modest effect (B = 0.01–0.04). The association was slightly attenuated in gender-diverse students compared with boys and girls.
Cross-sectional design prevents causal conclusions — it is unclear whether emotional distress drives AI-seeking or vice versa. The sample is limited to Ontario school boards, limiting generalizability. Full text was under embargo; findings are based on the abstract and results sections only.
When a young patient discloses using AI chatbots for emotional support, treat it as a clinical signal of psychological distress warranting further screening — independent of loneliness or social connectedness. Digital literacy programs and clinical frameworks should distinguish affective AI use from functional (school-related) use.
Explore related topics
