When Learning Processes Fail Novices: Technology Acceptance Through Mediation-Moderation Analysis in High School AI-Coding Education

This study examines an expanded Technology Acceptance Model (TAM) to evaluate the implementation of an AI and coding curriculum in high schools. Specifically, it integrates Perceived Usefulness and Perceived Ease of Use into Student Motivation as a mediator to analyze the relationship between Learning Processes and Learning Outcomes, moderated by Prior Coding Experience. Using an explanatory quantitative design, data were collected from 114 students at HelloMotion High School, a benchmarking school for national curriculum piloting. Structural Equation Modeling (SEM-PLS) revealed three critical findings. First, the learning process positively predicts learning outcomes (β = 0.244, p = 0.005). Second, student motivation functions as a significant partial mediator (indirect effect: β = 0.361, p < 0.001, which is stronger than the direct effect), confirming that the direct impact of pedagogy remains significant alongside psychological factors. Third, while the formal moderation of coding experience was statistically nonsignificant (p = 0.321), exploratory multi-group analysis showed that the learning process significantly impacted students with basic experience (p = 0.026) but failed to significantly impact complete novices (p = 0.160). These findings suggest that a "one-size-fits-all" curriculum is insufficient, highlighting the urgency for differentiated instructional strategies to support novice learners in national policy implementation.