Efficiency, Engagement, and Cognitive Offloading: A Review of Generative AI in Education

This narrative review examines how generative artificial intelligence (GenAI) influences student learning through three interrelated dimensions: efficiency, engagement, and cognitive offloading. A structured search of Scopus identified 56 studies published between 2022 and 2025 that examined GenAI use in educational contexts and its implications for cognitive or learning processes. The review reveals consistent evidence that GenAI enhances task efficiency by accelerating drafting, feedback cycles, and information processing. However, efficiency gains do not uniformly translate into deeper learning, as several studies report tendencies toward surface-level completion strategies. Findings related to engagement are mixed, with perception-based studies emphasizing reduced anxiety and increased confidence, while experimental studies report variable effects on cognitive involvement. Cognitive offloading emerges as a central tension: although GenAI use is associated with reduced cognitive load and support for higher-order reasoning, concerns persist regarding over-reliance and diminished independent thinking. To interpret these patterns, this review advances a three-lens framework positioning GenAI as a mediator of cognitive effort rather than a uniformly beneficial or detrimental tool. The framework highlights the interdependence of efficiency gains, engagement dynamics, and cognitive delegation. The findings underscore the critical role of learner regulation and instructional design in shaping GenAI’s educational impact. Future research should prioritize longitudinal designs, objective learning measures, and process-oriented methodologies to clarify long-term cognitive implications.