Working memory and generative AI tools in higher education: A systematic review

Article Sidebar

Download PDF
Published: Apr 16, 2026
DOI: https://royaliiteglobal.com/topic?id=400
Keywords:
generative artificial intelligence, working memory, cognitive offloading, critical thinking, higher education, cognitive load, metacognition

Main Article Content

Safar Bakheet Almudara
Department of Curriculum and Instruction, College of Education in Al-Kharj, Prince Sattam Bin Abdulaziz University, Saudi Arabia
Omar Abdullah Alshehri
Department of Library and Information Science, College of Education and Human Development, University of Bisha, Saudi Arabia
Swead Yahya Alzahrani
Department of Special Education, College of Education in Al-Kharj, Prince Sattam Bin Abdulaziz University, Saudi Arabia
Mohamed Sayed Abdellatif
Department of Psychology, College of Education in Al-Kharj, Prince Sattam Bin Abdulaziz University, Saudi Arabia

Abstract

Generative artificial intelligence (GenAI) tools have rapidly permeated higher education, yet their cognitive consequences for working memory, cognitive offloading, and higher-order thinking remain insufficiently synthesized. This systematic review, conducted in accordance with PRISMA 2020 guidelines, aimed to consolidate and critically evaluate empirical and theoretical evidence examining GenAI's effects on university students' cognitive functioning. A comprehensive search across six bibliographic databases covering 2023 to 2025 yielded 27 eligible peer-reviewed studies, appraised using design-matched quality instruments including RoB 2, AMSTAR-2, and MMAT. Findings reveal that unrestricted GenAI use measurably impairs long-term knowledge retention and critical thinking by substituting for the generative cognitive effort essential to durable memory encoding. However, structured, metacognitively scaffolded AI integration demonstrably preserves cognitive engagement and may enhance learning outcomes. Humanities students demonstrated disproportionate vulnerability to cognitive decline relative to STEM peers. Institutional policy frameworks, discipline-specific pedagogical redesign, and deliberate scaffolding strategies are identified as essential mitigating responses. The review concludes that cognitively informed, policy-guided GenAI integration represents the most defensible institutional approach, while urgent calls are made for longitudinal experimental research to establish stronger causal evidence.