Abstract
Retrieval-Augmented Generation (RAG) integrates information retrieval mechanisms with large language models (LLMs) in order to improve factual accuracy and contextual relevance in artificial intelligence systems. In educational environments, RAG-based systems can support intelligent tutoring, adaptive learning platforms, and reliable knowledge retrieval from large collections of academic materials. In this paper we review recent research on RAG in educational settings. We first present the definition of RAG, and briefly mention different types of retrievers and generation optimization methods. This study investigates the practical applications of RAG in educational contexts, with a specific focus on low-resource languages. Despite the growing body of literature on RAG in educational contexts, existing reviews reveal a notable limitation, the identified challenges and proposed future directions are predominantly oriented towards high-resource language environments. This paper addresses a critical gap in the field by systematically examining the comparatively sparse body of research pertaining to RAG applications in low-resource languages, an area that remains underexplored yet holds substantial implications for equitable and inclusive educational technology development.