ChatGPT Scaffolding in Supporting Metacognition for Limit Concepts in Guided Inquiry Mathematics Learning

Aim/Purpose: This study aims to investigate how ChatGPT-mediated scaffolding supports students’ metacognitive skills (planning, monitoring, and evaluating strategies) in understanding limit concepts in calculus within a guided-inquiry learning environment. Background: Guided inquiry fosters conceptual understanding in calculus, yet students often struggle with metacognitive regulation. While AI tools like ChatGPT offer interactive scaffolding, their impact on students’ self-regulated learning and problem-solving strategies in abstract topics, such as limits (a fundamental concept in calculus), remains underexplored. This study addresses this gap by evaluating ChatGPT’s function as a metacognitive guide in mathematics learning. Methodology: A convergent mixed-methods design was implemented with 75 students of mathematics education at Universitas Jambi over a period of four weeks. Participants engaged in guided inquiry activities on limits, using ChatGPT for problem-solving and reflection. Data was collected through pre- and post-metacognitive assessments, screen recordings of ChatGPT-student interactions, and reflective journals. Quantitative data were analyzed using paired t-tests, while qualitative data were thematically coded to identify patterns in metacognitive engagement. Contribution: This study advances understanding of AI’s capacity to foster self-regulated learning and critical thinking in mathematics, providing a framework for integrating generative AI as a metacognitive partner in guided inquiry pedagogy. Findings: Results indicate significant improvements in metacognitive skills, particularly in monitoring and evaluation strategies. Qualitative analysis revealed that ChatGPT’s iterative feedback encouraged students to critically analyze solutions, particularly in identifying boundary conditions in limit problems. However, 28% of students passively accepted AI-generated answers without deeper scrutiny, highlighting variability in engagement levels. Recommendations for Practitioners: Educators should integrate ChatGPT as a reflective tool in guided inquiry, designing structured activities that require students to justify or challenge AI-generated outputs. Providing explicit training in critical questioning techniques can enhance AI’s pedagogical value. Recommendation for Researchers: Future research should explore long-term retention of metacognitive skills developed through AI scaffolding and adaptive AI models for optimizing ChatGPT-student interactions in mathematics education. Impact on Society: The implications of this research extend beyond the classroom, potentially reshaping mathematics education in higher education. This approach could democratize access to personalized mathematical support, reduce educational inequalities, and prepare students for an AI-augmented professional landscape. However, careful consideration must be given to ethical implementation and the preservation of authentic mathematical thinking skills. Future Research: Further studies should examine (1) the sustainability of AI-enhanced metacognitive development, (2) cross-cultural differences in AI scaffolding effectiveness, and (3) improved AI-driven adaptive learning strategies for mathematics education.