Generative AI and Cognitive Processes in Education: A Narrative Review and Conceptual Framework

Generative artificial intelligence (GenAI) is increasingly being integrated into educational practices, transforming how students access information, develop ideas, and carry out cognitive tasks. While GenAI offers opportunities to foster critical thinking and creativity, its use raises concerns regarding "cognitive offloading"—the delegation of reasoning processes that are fundamental to learning. This study presents a review based on 108 documents from Scopus and Web of Science, which allowed for an examination of the relationships between GenAI, critical thinking, creativity, and cognitive offloading, as well as their pedagogical and institutional implications. The evidence analyzed suggests that GenAI can improve immediate task performance, facilitate idea generation, and support content development. However, these benefits do not necessarily translate into autonomous learning or greater creative originality. Outcomes appear to depend on students' prior knowledge, their AI literacy, task design, metacognitive regulation, and the nature and timing of pedagogical support. Based on this review, an AI-Mediated Cognitive Infrastructure Framework (AMCIF) is proposed, which links GenAI usage patterns with cognitive processes, educational practices, and institutional conditions. This framework distinguishes between forms of assistance that promote cognitive engagement and those that might replace the mental activity necessary for learning, emphasizing the essential role of instructor mediation, source verification, reflective practices, and process-oriented assessment. The AMCIF provides a conceptual foundation for analyzing how to integrate GenAI into educational settings while preserving critical judgment, creativity, and the capacity for independent reasoning. The proposed relationships require rigorous empirical validation.