This research examines the incorporation of generative AI technologies in education, emphasizing the creation of an AI-driven educational system that integrates retrieval-augmented generation (RAG), specialized fine-tuned large language models (LLMs), and asynchronous AI agents. The system seeks to improve educational results by providing precise, context-relevant answers customized to the exact requirements of each user, especially in Computer Science and Mathematics. Statistical analysis indicates exceptional accuracy (96.15%) and precision (97.22%) in the Computer Science domain. In Mathematics, the system achieves accuracy (83.34%) and precision (100%). Human assessments underscore clarity, specificity, and precision as primary strengths. Despite obstacles such as broadening domain coverage and maintaining data quality, the results demonstrate the system’s ability to promote adaptable learning environments.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Leveraging RAG and AI Agents to Enhance Specialized LLMs in Education

  • Laila El Jiani,
  • Sanaa El Filali,
  • Mohannad Tazi,
  • Abderrahmane Moujar

摘要

This research examines the incorporation of generative AI technologies in education, emphasizing the creation of an AI-driven educational system that integrates retrieval-augmented generation (RAG), specialized fine-tuned large language models (LLMs), and asynchronous AI agents. The system seeks to improve educational results by providing precise, context-relevant answers customized to the exact requirements of each user, especially in Computer Science and Mathematics. Statistical analysis indicates exceptional accuracy (96.15%) and precision (97.22%) in the Computer Science domain. In Mathematics, the system achieves accuracy (83.34%) and precision (100%). Human assessments underscore clarity, specificity, and precision as primary strengths. Despite obstacles such as broadening domain coverage and maintaining data quality, the results demonstrate the system’s ability to promote adaptable learning environments.