<![CDATA[The rapid integration of artificial intelligence (AI) in secondary education has transformed mathematics learning through adaptive and personalized instructional support. This study investigates the effect of AI-based learning on students’ mathematical thinking by comparing AI performance with student performance across different mathematical problem types. A quantitative experimental–comparative design with a pretest–posttest control group approach was employed, involving 120 secondary school students assigned to experimental and control groups. The experimental group engaged in AI-assisted learning using ChatGPT, while the control group received conventional instruction. Mathematical thinking was assessed using accuracy, response time, and consistency across multiple-choice, short-answer, and essay tasks, with digital literacy examined as a mediating variable. The results reveal significant performance differences, with AI demonstrating higher accuracy, faster response times, and greater consistency, particularly on essay-based problems requiring higher-order reasoning. Mediation analysis further indicates that digital literacy significantly influences students’ mathematical thinking and partially mediates the relationship between AI interaction and learning outcomes. These findings suggest that AI-based learning can effectively enhance mathematical thinking when integrated with strong digital literacy development, highlighting implications for adaptive pedagogy and future longitudinal research.]]>
