AI-Driven Remedial Learning for Advancing Learning Equity

• In Event: AI and EdTech for Peace and Inclusion: Accessing the Hardest to Reach Through Smarter Solutions

Proposal

This presentation will provide replicable frameworks for addressing educational marginalization across diverse contexts and offering practical strategies for incorporating peace-building principles into educational technology implementation worldwide. This study directly addresses the conference theme "Re-examining Education and Peace in a Divided World" by demonstrating how educational technology can mitigate divisions and promote social cohesion through inclusive access to quality learning. Particularly in conflict-affected and marginalized communities where educational exclusion perpetuates instability, regions experiencing conflict, displacement, and geographic isolation, traditional educational systems often fail to address educational needs of the most vulnerable groups, including children. If foundational learning access is not being provided, inequities that undermine social cohesion and peaceful development will continue to exist.

This RCT-validated education platform operates in both urban as well as rural settings, including diverse challenging contexts such as remote rural communities, and conflict-affected regions. With minimal infrastructure requirements (effective with one device per classroom) and operating entirely offline, the author is able to provide continuous access to its learning platform. The platform integrates structured pedagogy frameworks with comprehensive stakeholder engagement spanning students, teachers, education officers, and ministry officials. It also effectively utilizes AI-driven personalized learning algorithms that process millions of daily learning interactions for personalized learning experiences. All personal data is encrypted locally on devices using privacy-by-design approaches essential for data-sensitive contexts.
Specifically addressing these persistent challenges, the recently completed a pilot study examining Large Language Model-driven remedial interventions across 40 pre-primary schools in Kenya. The intervention used AI to generate personalized remedial lesson plans for struggling learners, with results showing students receiving the highest intervention dosage gained up to 1.765 standard deviations in emergent literacy compared to non-remedial groups. While promising, the study's small scale and limited duration require larger experimental validation to confirm these findings.
The author’s platform currently serves over 700,000 learners and 22,000 teachers across 11,000 schools in Kenya and Nigeria, with previous external evaluations demonstrating 0.534 standard deviations improvement in learning outcomes and cost-effectiveness analysis positioning it among the highest impact-per-dollar solutions globally. These findings demonstrate measurable progress toward educational equity and social inclusion essential for peacebuilding in divided societies.

Author

• Aidan Friedberg, EIDU