Between Code and Classroom: The Unequal Promise of Artificial Intelligence in Global Learning

Resumen

This study examines how artificial intelligence (AI)–powered personalized learning ecosystems shape educational equity and learning outcomes across four low-income countries in Sub-Saharan Africa and Southeast Asia, representing distinct levels of infrastructural and policy readiness. It explores the mediating role of pedagogical integration and the moderating effects of socioeconomic status and digital infrastructure on AI-enabled adaptive learning. Adopting a mixed-methods design, the research integrates large-scale learning analytics with qualitative insights from 60 interviews, 18 focus groups, and 24 classroom observations. Using Bayesian multilevel SEM and latent growth modeling, the study analyzes direct, indirect, and conditional effects of AI personalization on equity outcomes, supported by thematic analysis to contextualize implementation processes. Results reveal that AI-driven personalization improves learning equity when coupled with strong pedagogical integration and teacher facilitation. Learners from rural and low-income backgrounds show the largest gains under supportive socio-technical conditions. Cross-country comparisons emphasize the importance of policy commitment, infrastructure, and teacher digital competence. The findings propose a scalable framework for AI-enhanced learning equity, linking adaptive algorithmic design with human-centered pedagogy and inclusive educational development.