Self-Efficacy and Attitude on Trust in Artificial Intelligence Among Education Students: A Mixed-Method Study

Abstract

With the growing integration of Artificial Intelligence (AI) in education, it is crucial to understand how future educators perceive and interact with AI tools. This study examines how education students’ self-efficacy and attitudes influence their trust in AI, addressing the gap in existing literature that often treats trust as a mediator rather than an outcome variable. A mixed-method sequential explanatory design was used. In the quantitative phase, 140 education students answered standardized scales on AI self-efficacy, attitude toward AI, and trust in AI. Data were analyzed using descriptive statistics and regression analysis. In the qualitative phase, semi-structured interviews were conducted and analyzed through Interpretive Phenomenological Analysis (IPA). Findings revealed that students demonstrated moderately high AI self-efficacy, positive attitudes, and high trust in AI. Regression results showed that both self-efficacy (β = 0.605, p < .001) and attitude (β = 0.872, p < .001) significantly predicted trust, with attitude emerging as the stronger predictor. Qualitative themes included: (1) AI as Academic Scaffolding, (2) Negotiating Dependence and Reliability, and (3) Maintaining Agency and Control. Results highlight that students’ trust in AI is shaped by their attitudes and reinforced by their confidence in using it. AI is most effective as a supportive tool when paired with critical thinking and responsible use. The study recommends developing institutional guidelines for ethical AI use in the academe, grounded in AI literacy, ethical awareness, and self-regulation—ensuring that future educators engage with AI confidently, responsibly, and in alignment with SDG-4’s vision of inclusive and quality education.