Artificial intelligence use patterns, attitudes, and psychological distress among medical students: a cross-sectional survey

Abstract

Artificial intelligence (AI) tools are widely used by medical students, but associations of general use frequency and AI-related attitudes with psychological distress remain unclear. All 1,275 active students at one Turkish medical faculty were invited to an anonymous online survey conducted from 1 April to 1 May 2026; 249 of 251 respondents consented. Measures included general AI use, an investigator-translated 13-item Artificial Intelligence Attitude Scale (ATAI), and the validated Turkish Depression Anxiety Stress Scales-21. A general-distress composite was the primary outcome, and hierarchical regression used HC3 robust standard errors. AI use was reported by 95.6% of participants, and 50.2% reported daily use. Mean doubled DASS-21 total was 56.97 (SD 31.55). The ATAI three-factor model showed mixed fit (CFI = .898, TLI = .872, RMSEA = .107, SRMR = .070). AI-frequency categories did not improve the adjusted model at the block level (change in R -squared = .016, p  = .111; robust global p  = .088). Adding exploratory ATAI factors increased R-squared by.052 ( p  = .002; robust global p  = .009); Benefit was inversely, and Risk and Use positively, associated with distress. Subscale sensitivity analyses preserved coefficient directions, but only the inverse Benefit-Anxiety association survived correction across 15 additional focal tests ( q  = .032). AI use was nearly universal and served both educational and personal/non-educational purposes, but this sample did not provide stable evidence of an overall association between general use frequency and distress. ATAI associations are exploratory and may reflect measurement or common-method processes. Longitudinal multicenter studies should use formally adapted measures and objective indicators of AI engagement.