Advancing the Framework for Inclusive AI-Driven Personalized Learning: Multi-Phase Validation and Modeling Approaches

Artificial Intelligence (AI) offers strong potential to enhance personalized learning, but its implementation in low- and middle-income countries (LMICs) is limited by infrastructure constraints, ethical concerns, and lack of context-aware frameworks. This study proposes and validates an inclusive framework for AI-driven personalized learning aligned with Sustainable Development Goal 4 (SDG-4). A multi-phase methodology is used, combining expert consensus through the Delphi method with simulation-based modeling using large-scale educational datasets such as EdNet and OULAD. The findings highlight the importance of resilient architectures like Edge AI for low-connectivity environments and emphasize Expert-in-the-Loop governance to ensure culturally relevant, multilingual, and ethical AI deployment. The framework integrates pedagogical, technical, ethical, and inclusivity dimensions to support scalable and equitable AI-enabled learning in resource-constrained contexts.