Towards Intelligent Educational Resource Generation: A Survey of NLP, Deep Learning, and Large Language Models

Artificial Intelligence is transforming the way educational resources are designed, generated, personalized, and evaluated. Recent advances in Natural Language Processing (NLP), Deep Learning, Transformer architectures, Large Language Models (LLMs), Retrieval-Augmented Generation (RAG), and Multi-Agent Systems have enabled the automatic production of diverse educational materials, including courses, quizzes, slides, summaries, assessments, multimedia content, and personalized learning pathways. However, existing studies generally focus on isolated tasks such as question generation, intelligent tutoring, or educational chatbots, without providing a unified view of the educational resources that can be generated and the AI technologies that support them. This survey presents a comprehensive review of intelligent educational resource generation from both educational and technological perspectives. Following a structured survey methodology inspired by systematic literature review and systematic mapping principles, the paper synthesizes research from major scientific databases and classifies existing approaches according to educational resource type, AI methodology, personalization capability, and degree of generability. The survey introduces the concept of Educational Resource Generability, distinguishing fully generable, partially generable, and hardware-dependent educational resources. Furthermore, it proposes a unified taxonomy integrating educational objectives, resource modalities, AI technologies, personalization levels, and validation requirements. The paper also provides comparative analyses of NLP, Deep Learning, Transformer-based models, Large Language Models, Retrieval-Augmented Generation, and Multi-Agent Systems for educational content generation. Finally, it discusses current challenges—including factual reliability, hallucinations, pedagogical alignment, explainability, ethical considerations, multilingual support, low-resource languages, and multimodal generation—and identifies future research directions toward personalized, human-centered, and agentic educational ecosystems. This survey aims to serve as a comprehensive reference for researchers, educators, and practitioners interested in the next generation of AI-assisted educational resource generation