Cognitive mechanisms of AI agent behavioral intention among design graduate students: an integrated TAM–TTF perspective

AI agents are reshaping educational systems and transforming teaching and learning in design disciplines. However, the psychological mechanisms underlying AI-agent use among design graduate students remain underexplored, as does the interaction between design-specific tasks and AI agents. To address this gap, this study integrates the Technology Acceptance Model (TAM) and Task–Technology Fit (TTF) theory to explain how technological cognition and task-fit perceptions jointly shape behavioral intention to use AI agents. Because conventional structural modeling mainly captures hypothesized linear relationships, MLP-ANN was added to assess nonlinear patterns and predictor importance. Using 405 valid questionnaire responses, the model was tested through Partial Least Squares Structural Equation Modeling (PLS-SEM), a Multilayer Perceptron Artificial Neural Network (MLP-ANN), and Necessary Condition Analysis (NCA). PLS-SEM supported nine of ten hypotheses. Self-efficacy positively affected perceived usefulness (β = 0.458, p < 0.001) and perceived ease of use (β = 0.394, p < 0.001). Perceived enjoyment enhanced perceived ease of use (β = 0.220, p < 0.001), but its effect on perceived usefulness was negative and nonsignificant (β = −0.033, p = 0.483). Perceived usefulness (β = 0.376, p < 0.001), perceived ease of use (β = 0.185, p < 0.001), and task–technology fit (β = 0.224, p < 0.001) positively influenced attitude. Task characteristics (β = 0.322, p < 0.001) and technology characteristics (β = 0.266, p < 0.001) positively affected task–technology fit, while attitude promoted behavioral intention (β = 0.388, p < 0.001). Ten-fold cross-validated MLP-ANN models produced mean RMSE values of 0.3263–0.4203 across training and testing and broadly confirmed the PLS-SEM predictor-importance patterns. NCA further identified self-efficacy (d = 0.125, p = 0.015) and attitude (d = 0.132, p < 0.001) as necessary conditions for perceived usefulness and behavioral intention, respectively. This study extends the integrated TAM–TTF framework to design education and offers theoretical and practical implications for effective AI-agent use in design disciplines.