Large Language Model-Powered Personalized Education for Refugee Children: Adaptive Learning on Low-Resource Devices

Osama Orabi, Alaa Aldin Hajjar, Hadi Salloum

Track: Tiny paper

Keywords: LLM, Refugee, Education, Low-Resource Devices

TL;DR: This paper presents a low-resource framework that leverages large language models as adaptive personal tutors to mitigate educational deprivation among refugee children.

Abstract: Millions of refugee children suffer from prolonged educational deprivation due to the lack of formal schools, qualified teachers, and essential learning materials. This paper proposes a novel conceptual framework that leverages large language models (LLMs) to serve as personal tutors for each student. The proposed system adapts dynamically to individual learning styles and needs, while operating on low-resource devices commonly available in refugee camps. The framework includes data collection for personalized student embeddings, adaptive learning modules, lightweight local implementation, and human oversight through centralized monitoring.

Submission Number: 18