Regional Heterogeneity in College–Company Collaboration and Generative AI–Enabled Teaching in Vocational Undergraduate Education: Evidence from Multi-Group PLS-SEM

Abstract

Generative Artificial Intelligence (GAI) is increasingly used in vocational undergraduate education. Empirical evidence remains limited regarding how college–company collaboration, GAI-supported teaching processes, learning outcomes, and ethical risk perception are connected within vocational undergraduate education. This study developed and evaluated a structural model incorporating College–Company Collaboration Intensity (CCI), the AI teaching process (AI_Process), Learning Outcomes, and Ethical Risk Perception (ERP). The model was analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM) and Multi-Group Analysis (MGA). CCI had a significant positive effect on AI_Process (β = 0.742, p < 0.001). AI_Process showed a strong positive association with Learning Outcomes (β = 0.892, p < 0.001), while Learning Outcomes was positively associated with ERP (β = 0.635, p < 0.001). The structural model achieved a high explanatory power for Learning Outcomes (R² = 0.796), with AI_Process serving as the direct predictor. The multi-group analysis found differences among the three participating institutions in the CCI→AI_Process and Learning_Outcomes→ERP paths. The AI_Process→Learning_Outcomes path differed significantly in one pairwise comparison but not in the other two. The findings show that college–company collaboration was closely related to the integration of GAI into teaching, and that the AI teaching process was strongly associated with learning outcomes. Several structural paths differed across the three participating institutions.