Exploring AI Literacy Levels Among University Students Through the Lens of Self-Efficacy: A Case Study From a Historically Disadvantaged University

Abstract

Artificial intelligence (AI) is increasingly embedded in higher education, yet the role of psychological factors such as self-efficacy in shaping students’ readiness to use AI remains underexplored. This study examines how self-efficacy influences AI literacy levels, defined as students’ ability to understand, apply, and critically evaluate AI tools, among university students at a historically disadvantaged South African institution. Guided by Bandura’s self-efficacy theory, a quantitative survey was conducted with 153 students using a structured questionnaire measuring mastery experiences, vicarious experiences, social persuasion, and emotional states. Results show that more than 70% of students expressed confidence in understanding AI concepts, while 78% reported being able to learn new AI tools with ease. Patterns in the descriptive statistics further suggest that students who reported prior experience with AI, encouragement from peers and lecturers, and low levels of anxiety tended to express higher confidence in using AI. These findings indicate that self-efficacy is a critical enabler of AI literacy, with psychological readiness complementing technical competence. The unique contribution of this study lies in its focus on students from a resource-constrained, historically disadvantaged university and its integration of Bandura’s four self-efficacy dimensions into the study of AI literacy. The results suggest that interventions such as peer mentoring programs, hands-on AI workshops, and structured feedback sessions can enhance both confidence and competence, thereby supporting equitable AI adoption in higher education..