ALGORITHMIC BIAS AND ETHNOMATHEMATICS IN AI-BASED MATHEMATICS EDUCATION: A CRITICAL REVIEW AND INCLUSIVE PEDAGOGY

Abstract

The use of artificial intelligence (AI) in mathematics education expands opportunities for adaptive feedback and learning support but also raises concerns regarding algorithmic fairness, access, cultural representation, and student agency. This article aims to explain how these opportunities and risks are interconnected and to develop a pedagogical framework integrating ethnomathematics, critical mathematics education, and equitable AI governance. The study employs a critical integrative review of methodological, theoretical, empirical, and policy literature whose publication details can be verified. The literature is interpreted through four analytical dimensions: the learning benefits of AI, sources of inequity, recognition of culturally grounded mathematical knowledge, and conditions for implementation and governance. The synthesis indicates that the effects of AI cannot be generalized as invariably harmful or invariably beneficial. Inequities may arise when data, assessment design, device access, and pedagogical decisions disregard learners' contexts. Conversely, technology can support conceptual exploration when teachers retain instructional control and provide meaningful tasks. The article develops the Critical Techno-Ethnomathematics Pedagogical Framework (KPT-EK), comprising four pillars: algorithmic fairness auditing, dialogic cultural modeling, critical socioeconomic numeracy, and digital agency. The framework is operationalized through five instructional phases, multilevel evaluation indicators, and propositions for empirical testing. The article contributes a more balanced argument about AI and a testable pedagogical design. Its findings do not yet constitute evidence of the framework's effectiveness or a completed scoping review with an audited corpus of studies.