ChatGPT in education: A systematic review of its impact on critical thinking skills and dispositions

Abstract

Amidst the debate surrounding how generative AI like ChatGPT impacts critical thinking in classrooms, this study aims to address three questions: What does the existing evidence base look like? How exactly is ChatGPT being used in classrooms? And do these uses genuinely alter students' critical thinking skills and dispositions? We first review 11 existing syntheses and propose an analytical framework distinguishing between “skills” (Analysis, Inference, Evaluation, Deductive reasoning) and “dispositions” (Analyticity, Open-mindedness, Truth-seeking, Systematicity, Self-confidence, Inquisitiveness, Maturity), addressing the limitations of narrow research scope and lack of unified theory. Subsequently, following PRISMA guidelines, we systematically searched seven databases from December 2022 to October 2025. Applying strict inclusion/exclusion criteria and quality scoring, we retained only 65 high-quality empirical studies involving students. Using deductive + grounded open/axial/selective coding with NVivo, we identified five core application patterns in classrooms. Results indicate that teaching practices primarily embed ChatGPT within instructional workflows to cultivate students' analytical and evaluative skills, while relatively neglecting training in reasoning and deduction. While more frequently stimulating curiosity, openness, confidence, and structured thinking, systematic and long-term cultivation of deeper critical thinking dispositions—such as truth-seeking and mature judgment—remains lacking. This suggests that future teaching should evolve from merely “knowing how to critique” to fostering a willingness and responsibility to engage in critical thinking.