IMPACT OF CHATGPT USAGE ON CRITICAL THINKING AND ACADEMIC PERFORMANCE OF UNIVERSITY STUDENTS IN PAKISTAN

Abstract

The pervasive proliferation of generative artificial intelligence, most notably ChatGPT, within tertiary education has precipitated profound apprehensions concerning its ramifications for students’ cognitive maturation and scholarly probity. Although the extant corpus of scholarship comprehensively delineates the pedagogical affordances and perils associated with AI-augmented learning, rigorous empirical inquiry specifically addressing developing nations particularly Pakistan remains conspicuously deficient. The present investigation scrutinizes the interrelationship between the frequency and modalities of ChatGPT utilization and self-reported critical thinking competencies among Pakistani university students. Employing a quantitative, cross-sectional survey design, two psychometrically validated instruments were administered to a cohort of 86 students and 6 faculty members drawn from five public and private universities across Pakistan. Data were subjected to descriptive statistical analysis, Pearson product-moment correlation, and independent-samples t-tests utilizing SPSS v29. Findings indicate that the overwhelming majority of students (92.9%) reported regular weekly engagement with ChatGPT for academic purposes, while 43.0% acknowledged unedited reproduction (copy-pasting) of AI-generated material. Crucially, elevated frequency of ChatGPT utilization exhibited a statistically significant inverse association with critical thinking assessment scores (r = −.38, p < .01). Furthermore, while students overwhelmingly perceived ChatGPT as beneficial for their academic performance (84.9% reporting slight or significant improvement), objective indicators reveal substantial cognitive offloading and dependency, with 37.2% affirming a reduced need to think independently. Faculty respondents articulated substantially greater reservations than their student counterparts regarding these educational consequences, estimating significantly lower student copy-paste rates (15.0%) while overestimating source verification behaviors (70.0% vs. 53.5% actual). These empirical results underscore the urgent imperative for the institutionalization of structured AI literacy frameworks, the recalibration of assessment paradigms toward process-oriented evaluation, and the promulgation of coherent institutional policies that foster judicious, ethical, and responsible incorporation of ChatGPT within Pakistani higher education contexts