The Janus face of AI literacy: a dual-pathway model of its mitigating and exacerbating effects on academic procrastination

The rapid integration of generative artificial intelligence (GenAI) into higher education has created a paradoxical landscape, offering unprecedented access to knowledge while simultaneously posing risks for academic misconduct and procrastination. This dynamic is particularly salient in contexts lacking formal regulatory frameworks, where student behavior is primarily governed by individual psychological and competency-based factors. This study investigates the mediating role of AI literacy in the relationship between key psychological determinants reflected as perfectionism, academic integrity as well as academic self-efficacy and academic procrastination. Grounded in Bandura’s Social Cognitive Theory and Cognitive Load Theory, the research employs a cross-sectional survey design conducted from April to June 2025. Data from 654 Indonesian higher education students, recruited through quota sampling in an unmandated GenAI environment, were analyzed using SEM (SmartPLS). Validated instruments measured academic procrastination, AI literacy, perfectionism, academic integrity, and academic self-efficacy. Results indicate that perfectionism predicts procrastination, while academic integrity and self-efficacy are negative predictors. AI literacy partially mediates these relationships, demonstrating a dual role in mitigating and, under conditions, exacerbating procrastination. These findings highlight AI literacy not only as a technical competence but also as a psychological resource for managing cognitive load and supporting adaptive learning behavior in digital educational ecosystems.