<b>AfriML: A Culturally Relevant AI Tool for Teaching Machine Learning Concepts</b>

Abstract

This paper introduces AfriML, a culturally relevant artificial intelligence tool designed to enhance machine learning (ML) education through African cultural contexts. While tools like Teachable Machine and LearningML simplify ML concepts, they often lack cultural integration. AfriML addresses this gap by using artifacts from major Nigerian ethnic groups (Hausa, Igbo, Yoruba, Ibibio, and Efik) to teach ML classification problems. The tool was implemented with students at a tertiary institution and evaluated using a survey to assess learners’ confidence, satisfaction, and perceptions of cultural relevance. Results showed a significant increase in confidence scores from 2.43 to 3.28. Participants also reported improved understanding of ML classification (mean = 3.09) and rated the tool's effectiveness at 3.36. Cultural relevance was rated positively (mean = 3.45). These findings suggest that AfriML supports both conceptual understanding and cultural engagement in ML education. Future work will expand AfriML’s cultural scope and its impact across broader educational contexts.