Exploring Artificial Intelligence [AI]: Understanding Relationships between AI Use and Mental Health Among Middle and High School Students

Key Findings

• — 81% of students reported using any AI; 58% used it for schoolwork, 28% for relationship or problem-solving advice and 12% for friendship or companionship
• — 57% used AI when they felt sad, stressed, lonely or overwhelmed, but only 8% said yes when asked directly whether they used AI to get help at such times
• — Students who used AI for any reason did not significantly differ from non-users on mental health, well-being or suicide indicators (adjusted prevalence ratio for lifetime suicidal thoughts 1.04, 95% CI 0.88-1.22)
• — Using AI as a friend or companion was associated with lifetime suicidal thoughts (aPR 2.01, 1.78-2.26) and attempts (aPR 1.98, 1.58-2.49); seeking help from AI when emotionally struggling was associated with suicidal thoughts (aPR 2.43, 2.15-2.75) and attempts (aPR 2.70, 2.14-3.40)
• — Among reasons for using AI when distressed, the strongest associations with lifetime suicidal thoughts were 'I don't have anyone to talk to' (aPR 2.85, 2.53-3.22) and 'when I needed to talk but no one else is around' (aPR 2.60, 2.32-2.92)
• — Middle school students were more likely than high school students to use AI when feeling sad, stressed, lonely or overwhelmed (65% vs approximately 50%); females more than males (62% vs 51%), and females more often used AI for friendship or companionship (15% vs 9%)
• — American Indian/Alaska Native (72%), Black (65%), Hispanic (65%) and multiracial (56%) students were more likely than white students (47%) to use AI when emotionally struggling

Methodology Notes

Cross-sectional analysis of JED's Student Survey (online, school-administered with parental consent and teen assent) from 21 schools in nine high schools and districts that joined JED programmes in the 2025-26 academic year; data collected October 2025 to April 2026; n=5,560 after inclusion criteria. Weighted for non-response to school/district enrolment characteristics (NCES data). Outcomes: PHQ-4 (anxiety and depression), UCLA Loneliness Scale, Diener Flourishing Scale, Brief Resilience Scale, lifetime suicidal thoughts and attempts (binary). Generalized mixed additive models accounting for nesting, adjusted for sex, race/ethnicity and grade; significance after false discovery rate adjustment. AI-use items developed with experts (not a validated instrument). The authors state the sample is not nationally representative and causality and direction cannot be inferred. Observation dates are month precision (October 2025 to April 2026). Funded by the Nielsen Foundation, with input from the Digital Wellness Lab at Boston Children's Hospital. Not peer-reviewed. Verification: landing page, press release and 40-page PDF fetched 2026-10-29.

Sources