Generative AI Literacy and Research Self-Efficacy as Predictors of Attitudes Toward Scientific Research Among Health Sciences Students: A Cross-Sectional Study

Generative artificial intelligence (GenAI) has rapidly permeated undergraduate research training, yet its relationship with research self-efficacy and attitudes toward scientific research remains underexplored in health sciences students. This cross-sectional study examined how GenAI literacy and research self-efficacy relate to attitudes toward scientific research, and whether their contribution is direct or channelled through self-efficacy. A total of 91 undergraduate health sciences students enrolled in research courses at a private Peruvian university with campuses across several regions completed a 36-item Likert scale covering three higher-order constructs, validated by three expert researchers. Data were collected between March and May 2026. Analyses combined descriptive statistics, Spearman correlations, HC3 robust multiple regression, and fuzzy-set qualitative comparative analysis (fsQCA). Reliability was adequate to excellent (GenAI literacy α = 0.861, AVE = 0.675; research self-efficacy α = 0.927, AVE = 0.810; attitudes α = 0.790, AVE = 0.603). GenAI literacy correlated with attitudes (ρ = 0.475, adjusted p < 0.001) but lost independent weight once self-efficacy entered the regression (b = 0.084, p = 0.570), whereas research self-efficacy remained the strongest adjusted predictor (b = 0.356, 95% CI 0.091-0.620, p = 0.009); the model explained 40.4% of the variance (F(4, 86) = 14.60, p < 0.001). Discriminant validity between GenAI literacy and self-efficacy was incomplete (HTMT = 0.912). The fsQCA identified functional knowledge and critical-ethical evaluation of GenAI as core conditions present in every sufficient configuration for high attitudes (intermediate solution consistency = 0.913). Findings indicate that GenAI literacy shapes research attitudes indirectly, through research self-efficacy, and that its ethical-critical dimension is a recurrent ingredient of favourable dispositions. Interventions should strengthen research confidence and critical GenAI literacy jointly rather than as separate competencies.