The Generative AI Cognitive Effort Paradox: When Less Effort Does Not Mean Better Learning – An Integrative Review of Cognitive Load, Productive Struggle, and Independent Learning

Generative artificial intelligence (GenAI) is reshaping higher education by reducing the time and effort required to access information, solve problems, and complete academic tasks. Although these capabilities can improve learning efficiency, their implications for deeper understanding and independent learning remain uncertain. This review examines the cognitive effort paradox in GenAI-supported education: the possibility that reducing unnecessary mental effort can facilitate learning while excessive cognitive delegation may weaken the processes through which knowledge is constructed and retained. Drawing on Cognitive Load Theory, desirable difficulties, self-regulated learning, and cognitive offloading research, the review distinguishes productive cognitive assistance from cognitive substitution. It examines how GenAI influences cognitive load, learning engagement, metacognitive monitoring, knowledge retention, and independent problem-solving. The synthesis suggests that the educational value of reduced cognitive effort depends on the nature of the task, learner characteristics, instructional design, and whether students remain actively involved in learning-relevant processing. An integrative conceptual framework is proposed to explain the conditions under which GenAI-supported efficiency may strengthen or constrain learning. The review offers directions for experimental research, educational assessment, and responsible GenAI integration in higher education.