🧠 AI Ed Wiki

🏷️ Concept

Automated Question Generation

● high · created 2026-05-08 · updated 2026-05-22nlp-education, assessment, llm, stem-education, automated-grading, higher-ed, generative-ai, intelligent-tutoring, student-experience, scaffolding

Automated question generation leverages NLP and LLMs to create educational assessments at scale. Wei & Stamper (2025) introduced the generate-then-validate paradigm, reducing hallucination by 62% compared to direct generation and achieving 89% accuracy on STEM datasets.

Generate-Then-Validate Paradigm

1. Generation Phase: LLM produces candidate questions from source material

2. Validation Phase: Separate verification step filters invalid/low-quality items

3. Refinement Loop: Failed items trigger re-generation with corrective prompts

Advantages Over Direct Generation

  • Reduced Hallucination: Validation catches factually incorrect questions
  • Higher Relevance: 23% improvement on relevance metrics vs. baseline LLMs
  • Scalability: Enables rapid creation of formative assessments across domains

    • Educational Applications

  • Formative Assessment: Just-in-time questions for adaptive learning systems
  • STEM Education: Validated on mathematics and science problem generation
  • Differentiated Instruction: Generating multiple difficulty levels automatically

    • References

    Wei, Y., Stamper, J., et al. (2025). Generate-Then-Validate: A Novel Question Generation Approach. arXiv:2512.10110.

    Source

  • https://arxiv.org/abs/2512.10110

    • Connected Concepts

  • Personalized Learning
  • Human In The Loop AI
  • Edtech Platform
  • Learning Analytics
  • LLM
  • Intelligent Tutoring
  • Scaffolding
  • Pedagogical LLM Training
  • Formative Assessment
  • Adaptive Learning
  • Automated Assessment
  • RAG

    • Connected Articles

  • Code Gen
  • Generate Then Validate Question Gen
  • Kt4eqg Personalized Question Generation
  • Nsmq Riddles Science Math Benchmark
  • Self Referential L2 Writing LLM Assessment
  • Slidesqaqa Pedagogical Question Generation
  • Taklif AI Interest Based Personalized Assignments
  • Zerkouk Comprehensive Review ITS 2025