Advancing Medical Education in Sex- and Gender-Based Mechanisms of Disease: Current Educator Knowledge and Future Directions.

With the rapid progress of artificial intelligence (AI), the technology is being increasingly adopted in mathematics education. Applications like ChatGPT, DeepSeek, and Seewo Whiteboard create new opportunities for personalized and interactive learning, instant feedback, and the cultivation of students’ mathematical potential (; ; ; ; ). However, many studies have shown that large language models (LLMs) present obvious reasoning limitations at all levels of mathematics, from primary school problems to International Mathematical Olympiad and advanced mathematics (; ; ; ; ). Some studies have also indicated that LLMs do not ensure that the tutoring support is either valid or pedagogically appropriate, such as deviating from expected pedagogical strategies or leaking answers to students (). As points out, we need to carefully consider two risks of LLMs in mathematics education. First, because they lack genuine symbolic reasoning, they frequently cause significant mathematical errors. Second, they remain incapable of applying suitable teaching strategies. Therefore, whether AI can indeed play a positive and effective role in mathematics teaching largely depends on teachers’ willingness and ability to critically use such tools. For example, teachers need to evaluate AI outputs for mathematical accuracy, pedagogical appropriateness, and alignment with student thinking (; ; ; ). Unlike general technology acceptance, critical behavioral intention (CBI) involves a reflective, evaluative orientation, which is particularly important in the professional contexts of mathematics teaching. Specifically, CBI refers to a reflective and evaluative stance toward AI in mathematics, characterized by critically assessing AI outputs for mathematical accuracy and pedagogical appropriateness while maintaining independent judgment. Moreover, teachers cannot blindly trust AI, rather, they must adopt a critical stance, and their ability to do so depends on their own professional knowledge. According to , whether pre-service teachers can formulate effective prompts with GenAI depends on their Intelligent-TPACK, a domain-specific knowledge construct. This suggests that professional knowledge shapes AI-mediated task performance. found that higher dependency on generative AI among students was significantly associated with increased cognitive failures and weakened critical thinking. This suggests that excessive reliance on AI can weaken the cognitive abilities essential for mathematics learning, such as logical reasoning and independent judgment. This finding serves as an important reminder for mathematics teachers that they must prioritize cultivating their own CBI, and then guide students to also develop their habit of using AI critically. From this perspective, as noted by our research team, AI essentially maps the static body of human mathematical knowledge (e.g., published papers, textbooks, monographs) onto a dynamic knowledge system, akin to a functional mapping. In teaching and self-learning, static knowledge can be conceptualized as nodes in a grid; when using AI, teachers and students pose questions around these nodes, and AI progressively refines and completes their thinking. Consequently, in the AI era, rote memorization becomes less meaningful, whereas understanding, judgment, and logical reasoning become paramount. It is precisely in this dynamic process, moving from static knowledge nodes to active questioning and refinement, that students’ mathematical potential can be tapped. This further underscores why teachers’ critical behavioral intention is essential.