Latent profiles of generative artificial intelligence use among Chinese college students: Associations with depression and anxiety

Key Findings

• — Four usage profiles identified by latent profile analysis; the Problem-Dependent profile (15% of students; high escapism, low AI literacy, high dependency) showed the highest depression and anxiety
• — The Rational-Tool profile (high AI literacy, instrumental motivation) showed the most favorable mental-health outcomes
• — Profiles were recoverable by a random-forest classifier from correlates not used to build them (accuracy 0.81, macro-average AUC 0.89), with Problem-Dependent the most distinguishable (AUC 0.94)
• — SHAP analysis identified depression, smartphone addiction, and executive-function difficulties as key correlates of the Problem-Dependent profile
• — Authors argue prevention should target usage motivation, AI literacy, and dependency rather than usage frequency alone

Methodology Notes

Cross-sectional survey (n=5,748), Northwest Normal University (Lanzhou) team; BDI-II and BAI symptom measures; latent profile analysis plus an exploratory random-forest/SHAP classification. Online-ahead-of-print 2026-08-27 (PMID 42660392); Crossref print date is month-precision 2026-08. ScienceDirect blocks automated fetchers, so verified via the full PubMed-deposited structured abstract plus Crossref.

Sources