The Impact of Artificial Intelligence on Students' Critical Thinking: A Systematic Synthesis, Conceptual Model, and Comparative Review

Abstract

The rapid diffusion of AI-based applications such as ChatGPT in academic and self-study environments has further stimulated debates on the effects of such applications on the development of critical thinking skills among students. This review will be a systematic review of the existing literature on this topic published from 2011 to 2026, both theoretical and experimental. Building on the cognitive load theory, self-regulated learning theory, Bloom's taxonomy revision, and cognitive offloading research, this paper develops an original Critical Thinking Paradox Model, which models both ways in which AI influences critical thinking through a reliance-dependent mathematical function. According to the synthesis, AI applications are effective at enhancing academic performance, motivational-affective state, and lower-order cognitive processes, but the evidence concerning higher-order cognitive processes, metacognition, and autonomous reasoning is inconsistent and dependent upon scaffolding, tasks design, and students' AI skills. Neurophysiological data suggests reduced neural connectivity and decreased feeling of ownership among students writing essays without scaffolding from large-language models compared to unsupervised and search-engine-based essay writing. A comparative review of 15 representative studies is presented, along with five original figures and six mathematical formulations. Based on these findings, the paper provides practical recommendations for promoting AI literacy while preserving students' critical thinking skills in AI-supported educational environments. It also outlines key directions for future research, including longitudinal studies, cross-cultural investigations, and neuroscientific research.