PEDAGOGICAL STRATEGIES FOR DEVELOPING CRITICAL THINKING AMONG PRE-SERVICE TEACHERS THROUGH THE USE OF GENERATIVE ARTIFICIAL INTELLIGENCE

Abstract

The integration of generative artificial intelligence into higher education has raised questions about its relationship with students' critical thinking and the pedagogical conditions under which these technologies are used. This study examines the associations between reflective GenAI use, perceived pedagogical guidance, and self-reported critical thinking competencies among pre-service teachers in Kazakhstan. A quantitative cross-sectional survey was conducted between September and October 2026, involving 293 students from five higher education institutions. The research instrument comprised 20 items organized into five scales measuring reflective GenAI use, passive reliance on AI-generated responses, digital competence, pedagogical guidance, and critical thinking. Data were analysed using descriptive statistics, Cronbach's alpha, Spearman's rank-order correlations, and ordinary least squares regression. The scales demonstrated acceptable internal consistency (α = 0.803–0.856). Reflective GenAI use was positively associated with self-reported critical thinking (ρ = 0.542, p < .001), while perceived pedagogical guidance was positively associated with reflective AI-use practices (ρ = 0.427, p < .001). Both relationships remained statistically significant after adjustment for selected individual characteristics. Passive reliance on AI-generated responses was negatively associated with critical thinking (ρ = −0.311, p < .001), whereas the frequency of GenAI use was not a significant independent predictor in the adjusted regression model. The findings suggest that students' reported critical thinking is more closely associated with the cognitive strategies accompanying AI use than with its frequency. The study contributes to research on AI-supported teacher education by distinguishing reflective engagement from passive reliance and identifying perceived pedagogical guidance as a correlate of students' evaluative learning practices. The results support the consideration of instructional approaches that encourage independent verification, critical evaluation, and reasoned interpretation of AI-generated information.