Modeling the Determinants of AI Adoption in Authentic Online Assessment: An Integrated TPB–TAM Framework for Teachers in Technology-Enhanced Learning Contexts

Abstract

In the current era where artificial intelligence technology plays an increasingly important role in education, teachers are increasingly interested in applying AI to enhance learning efficiency and assessment. However, the acceptance of AI in assessment remains diverse, both helping to make education more equal and effective. At the same time, some are concerned that AI may replace the role of teachers or cause negative impacts. This study aimed to create a causal model explaining the determinants of the use of artificial intelligence (AI) in assessing real-world online learning outcomes of teachers in basic education by integrating the conceptual frameworks of the Technology Acceptance Model (TAM) and the Theory of Planned Behavior (TPB). It covered both technology perception factors, namely Trust in AI, Barriers to AI Adoption, Technology Self-Efficacy, and planned behavioral factors, namely Attitude Toward Behavior, Subjective Norms, and Perceived Behavioral Control, to predict teachers’ behaviors to accept AI in real-world online assessments. The sample consisted of 260 basic education teachers, selected by multi-stage random sampling in schools that used online assessments. A five-point scale questionnaire was employed as a research tool which was tested for content validity and internal reliability. Structural Equation Modeling (SEM) was used as data analysis. The results showed that the model demonstrated excellent fit indices (GFI = 1.000, AGFI = 0.997, RMSEA = 0.000), and explained 79.1% of the variance in AI adoption behavior (R² = 0.791). The proposed causal model could explain the variance in AI usage behaviour significantly, where the variable of AI adoption in teachers’ real-world online assessment (AAB) was directly influenced by the variables of attitude toward AI use in assessment (ATB), social norms (SN), perceived behavioral control (PBC), AI trust (TA), and technology self-confidence (TSF), all of which were statistically significant. In addition, the high barriers to AI use had a negative effect, indicating that teachers were less likely to adopt AI in real-world online assessments. This finding indicates that teachers make rational decisions to accept technology based on perceived value, rather than social pressure. The promotion of AI should focus on developing teachers’ knowledge and skills, along with creating a supportive environment that reduces the difficulty of using such technology, and avoiding direct enforcement through orders or regulations. Future research should explore longitudinal trends and include contextual or institutional variables that may affect teachers’ decision-making regarding AI use.