Panorama of AI and LLM Applications in Education: In-depth Interpretation of the awesome-ai-llm4education Resource Library

Basic Information of the Resource Library

• Maintainer: GeminiLight
• Source: GitHub
• Core Value: Systematically organizes the latest research in the field of AI and LLM applications in education, covering eight major areas such as intelligent tutoring and learning path recommendation
• Role: Provides a literature navigation tool for educational technology researchers, AI developers, and frontline educators

This resource library is continuously updated (since 2024) and includes papers from top conferences (KDD, CHI, etc.), journals, and preprints, addressing the pain point of scattered research.

Artificial intelligence (especially LLMs) is profoundly transforming education, but the field is developing rapidly, with research scattered across various conferences and journals, making it difficult for researchers to fully grasp the latest progress.

GeminiLight's awesome-ai-llm4education resource library emerged as a solution, aiming to systematically collect and organize relevant papers to provide clear knowledge navigation for practitioners.

The resource library divides AI education research into eight clearly layered fields:

1. Overview, Analysis, and Vision: Macro-perspective reviews and trend analysis
2. Tutoring Strategies: Learning path recommendation, intelligent tutoring systems
3. Learning Experience: Engagement enhancement, student profile construction
4. Assessment and Feedback: Adaptive testing, automatic scoring, cognitive diagnosis, knowledge tracing
5. Teaching Material Preparation: Content generation, knowledge structuring, question generation
6. Teaching Assistance: Teacher lesson planning and classroom management assistance
7. Specific Scenario Applications: Vertical fields by subject (computer science, language, mathematics, etc.)
8. Datasets and Benchmarks: Standard datasets and evaluation benchmarks for educational AI tasks

Innovations in Learning Path Recommendation

• Multi-Agent Learning Path Planning via LLMs (Xu et al., 2026): A planning paradigm combining multi-agent systems and LLMs
• LearnMate (Wang et al., CHI 2025): An interpretable personalized learning system
• PlanGlow (Chun et al., 2025): A learning planning approach emphasizing user controllability

Evolution of Adaptive Testing and Cognitive Diagnosis

• Survey of Computerized Adaptive Testing (Liu et al., 2024): A review from the machine learning perspective
• Shift from traditional IRT to deep knowledge tracing models (combining Transformer/graph neural networks)

Paradigm Shifts Brought by LLMs

• Zero-shot/few-shot capabilities cover diverse educational tasks
• Challenges: Hallucination issues, insufficient fine-tuning on educational expertise, high computational costs

• Educational Technology Researchers: Quickly locate the latest progress and identify research gaps
• AI Developers: Understand special needs of educational scenarios (data sparsity, interpretability, ethical privacy)
• Frontline Educators: Gain inspiration for integrating AI tools into teaching
• Policymakers: Grasp the trends of AI education applications to assist decision-making

Future Trends

• New technologies such as multimodal large models and embodied intelligence will expand application scenarios
• Fairness, transparency, and security will become important research topics

Learning Recommendations

1. Start with review papers to build an overall understanding
2. Dive deep into specific subfields based on interests
3. Follow the continuous updates of the resource library to keep up with domain dynamics

This resource library is not just a collection of papers, but also outlines the panorama of AI education research, helping practitioners find their way in the ocean of papers.