An Adaptable Client-Server Architecture for Generating Educational Content Using Large Language Models

Recent advancements in generative artificial intelligence (GenAI), particularly large language models (LLMs), have transformed the landscape of AI-driven educational applications. In this paper, we report on the design and use of a general and adaptable client-server web application architecture that harnesses LLMs for automated educational content generation. This architecture seamlessly integrates modern web technologies with AI-driven content creation workflows, enabling instructors to generate instructional materials and assessment items efficiently. The system leverages retrieval-augmented generation (RAG) to incorporate relevant course materials, ensuring that generated content aligns with predefined learning objectives and pedagogical frameworks. Additionally, prompt engineering techniques are employed, leveraging structured course modeling, and human-AI interaction in optimizing the quality and usability of AI-generated content. To evaluate the effectiveness of this architecture, we discuss the outcomes of multiple research studies that implement this framework in a research setting. These studies examine various use cases, AI integration strategies, and iterative improvements in content generation, highlighting both the potential and challenges of LLM-driven educational applications. Furthermore, the application of this architecture to real-world educational settings is discussed. By providing a scalable, adaptable, and research-driven approach, this work contributes to the ongoing development of AI-enhanced learning environments, paving the way for future innovations in automated content generation, adaptive learning, and AI-assisted instruction.