Artificial intelligence is increasingly entering mathematics classrooms through tools for feedback, personalization, assessment support, and instructional decision-making; however, its potential to contribute to inclusion depends less on technical availability than on how teachers conceptualize what AI is, what it can legitimately do, and what risks it introduces. The discussion is framed primarily around K–12 mathematics classrooms (primary and secondary) and targets both pre-service preparation and in-service professional development for mathematics teachers. This Conceptual Analysis clarifies the construct of teachers’ conceptions of AI for inclusive mathematics learning and argues that such conceptions shape not only adoption decisions but also the quality and equity of classroom use. In a first approach, the paper delineates analytically distinct dimensions of conceptions, including beliefs about AI capabilities and limits, professional role and identity, ethical governance concerns, and perceived institutional conditions. Then, it outlines a minimal typology showing how different configurations of these dimensions can generate predictable inclusion-related barriers, such as over-automation of pedagogical judgement, opacity in feedback and decision processes, and unequal exposure to risk in vulnerable contexts. Later, it synthesizes governance-linked conditions for informed engagement, making explicit the safeguards under which AI-supported practices are more compatible with inclusion and the criteria under which cautious non-adoption is professionally warranted. The analysis culminates in actionable implications for teacher education, foregrounding the capabilities required for accountable mediation of AI-supported practice in mathematics classrooms, including boundary-setting, bias-awareness, transparency, and context-sensitive orchestration.
Niño et al. (2026) studied this question.