Structural links between AI-art literacy, VR presence, visual-language proficiency, and experimental imaging performance among university students.

Abstract

This paper examines the emerging digital literacies and immersive technology and their influence on real-world making in university studios. We modeled our hypothesis and tested it on cross-sectional data (n = 476 students in Beijing, Shanghai, Shenzhen, Hangzhou, Chengdu and Wuhan) where AI-art literacy predicts performance in experimental imaging, with visual-language proficiency mediating and VR presence moderating the relationship. We resorted to various measures: self-reported, validated and short knowledge tests (including Mini-VLAT-like items), and we scored performance using a rubric complemented with system log information. Analyses in SmartPLS 4 (5000 bootstraps) confirmed strong reliability/validity and supported the structural model: AI-art literacy positively predicted experimental imaging performance (β = 0.14, p = .002) and visual-language proficiency (β = 0.62, p < .001); visual-language proficiency substantially predicted experimental imaging performance (β = 0.45, p < .001), yielding a significant indirect effect (β = 0.28, p < .001). VR presence both moderated the proficiency → performance path (β = 0.10, p = .008) and exerted a direct effect on performance (β = 0.16, p < .001). Demographic and background covariates were uniformly non-significant. The model explained R2 = 0.39 of visual-language proficiency and R2 = 0.52 of experimental imaging performance. Novelty lies in a production-centered, theory-consistent pathway from domain-specific AI-art literacy to an authentic, rubric-and-logs outcome via visual-language proficiency, with VR presence as a boundary condition, delivered through a reflective-formative PLS-SEM architecture and a large, multi-city Chinese sample.