From tool to learning partner: how generative AI shapes perceived learning support among design students

As generative AI tools are increasingly adopted in design education, understanding how AI learning assistants foster perceived learning support (PLS) among design students has become a pressing concern. Drawing on Social Support Theory, the Computers Are Social Actors (CASA) framework, Parasocial Relationship Theory, and the Stimulus–Organism–Response (SOR) model, this study examines how AI anthropomorphism and technical quality factors shape trust and enjoyment, ultimately influencing PLS. Using a multi-method quantitative approach, PLS-SEM (N = 297), Artificial Neural Network (ANN), and Necessary Condition Analysis (NCA), we find that perceived enjoyment (β = 0.543, p < 0.001) and AI trust (β = 0.320, p < 0.001) are the strongest direct predictors of PLS, while technical quality factors act indirectly through enjoyment and trust. Anthropomorphism shows a paradoxical pattern: human-like features enhance enjoyment and trust, yet have a small negative direct effect on PLS (β = − 0.119, p = 0.002), suggesting that overly human-like responses may raise expectations for authentic support that the AI system cannot fully meet. The model explained 64.5% of the variance in PLS with an acceptable fit. This study advances understanding of student–AI learning partnerships in design education and offers design insights for calibrating anthropomorphic features to foster sustainable learning support.