Answers are grounded in this paper and cite their sources.
The utilization of AI chatbots in mathematics education is on the rise, with these bots often serving as the primary providers of answers, potentially limiting learners’ engagement in reflective and metacognitive reasoning processes. The present study examines how AI chatbots can function as metacognitive partners within a culturally contextualized mathematics learning framework implemented through a learning management system (LMS). Using an embedded multiple-case study design, the data were collected from 33 preservice teachers enrolled in the Department of Mathematics Education during their seventh semester. The data were collected from several sources: chatbot interaction logs, participatory observations, metacognitive reflective journals, and semi-structured interviews. Four participants were purposively selected for in-depth analysis. Thematic analysis revealed two distinct patterns in the interaction of pre-service mathematics teachers with the AI chatbot when completing assignments: an iterative interaction pattern and a linear interaction pattern. The iterative pattern involves repeated exploration, whereas the linear pattern follows a straightforward path with minimal verification. The two patterns yield different metacognitive outcomes, depending on the degree of regulation. When iterative interaction is well regulated, it fosters metacognitive success; however, when it becomes overly regulated, it leads to a metacognitive failure known as mirage, in which learners overanalyze and begin to question correct reasoning. Similarly, when linear interaction is well regulated, it also facilitates metacognitive success. Yet, when it becomes less regulated, it results in a metacognitive failure, blindness, due to the uncritical acceptance. The findings of the present study suggest that AI-supported learning functions as a process of co-regulation rather than passive information delivery, underscoring the importance of regulatory awareness in culturally mathematics learning.
No takes yet. Share an insight, caveat, or question.
Alifiani et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: