GENERATIVE AI-ASSISTED ACADEMIC WRITING AND CRITICAL THINKING AMONG PAKISTANI UNIVERSITY STUDENTS: A MIXED-METHODS INVESTIGATION

Abstract

The rapid integration of generative artificial intelligence (GenAI) into higher education has transformed academic writing practices while raising concerns about its implications for students' critical-thinking development. This study investigated the relationship between GenAI-assisted academic writing and critical thinking among university students in Pakistan, with particular attention to whether GenAI functions as a cognitive scaffold or a substitute for independent reasoning. A sequential explanatory mixed-methods design was employed, comprising a quantitative survey followed by qualitative interviews. The quantitative phase examined AI-assisted idea generation, drafting and organization, language and revision, argument development, feedback and evaluation, and AI evaluation and verification in relation to critical thinking. The qualitative phase explored students' experiences of productive AI use, critical evaluation, and cognitive dependence. The findings indicated significant positive relationships between GenAI-assisted writing dimensions and critical thinking, with AI evaluation and verification emerging as the strongest predictor (β = .39, p < .001), followed by AI-assisted feedback and evaluation (β = .22, p < .001) and argument development (β = .19, p < .001). The qualitative findings further indicated that GenAI supported critical thinking when students used it for feedback, alternative perspectives, argument refinement, and verification, whereas excessive delegation of cognitive tasks encouraged dependence and reduced independent reasoning. The findings extend Cognitive Load Theory by distinguishing between productive cognitive-load reduction and cognitive offloading. The study concludes that the educational value of GenAI depends less on its availability than on how critically and purposefully students interact with it. Universities should therefore emphasize AI literacy, source verification, reflective AI use, process-based assessment, and responsible human–AI collaboration.