Integrating Artificial Intelligence into STEM Education: A Study on Technology Acceptance, Ethical Concerns, and Disciplinary Differences Among Teachers in Macau

Abstract

This study investigates AI acceptance, ethical concerns, and disciplinary differences among 152 STEM teachers and 673 students in Macau secondary schools. Building upon the Technology Acceptance Model (TAM), the study incorporates two new constructs—Subject Specificity (SS) and Cross-disciplinary Integration (CI)—to develop an extended TAM model. Path analysis (SEM) revealed a three-tier progressive model: ease of use as foundation (PEOU, M=3.55), usefulness as core driver (PU, M=3.79), and disciplinary fit as ultimate determinant (SS, M=3.71). Subject Specificity showed significant direct effects on Perceived Usefulness (β=0.715) and Behavioral Intention (β=0.610), with PU mediating PEOU and BI. Future studies will expand SS/CI and ethics items to enhance construct validity. Teachers scored significantly higher than students on PU (t=2.15, p<.05, d=0.20). Among ethical dimensions, Academic Integrity (M=4.20) scored highest and Role Transformation (M=3.28) lowest. Three teacher groups emerged: Neutral Acceptors (52.6%), Effect-Driven (11.2%), and High Intention but Low Confidence (17.1%). Student High Adopters (23.5%) far exceeded teachers (10.5%), suggesting a "students knowing AI better" phenomenon. The findings provide empirical evidence for targeted AI education policies and differentiated teacher training programs.