Designing Classroom Assessments with Generative AI
Guher Gorgun, Okan Bulut, Bin Tan · 2025
This chapter provides a practical guide on automatic question generation using large language models (LLMs) within a teacher-in-the-loop framework. We begin the chapter with an overview of current methods used for automatic question generation and highlight several challenges that may impede educators from fully leveraging artificial intelligence (AI) tools for question generation. Next, we present a three-stage framework that facilitates generating questions using LLMs. In the first stage, educators create a blueprint highlighting the knowledge and skills covered in class. In the second stage, the blueprint guides the prompt engineering process, where teachers develop detailed and descriptive prompts to generate questions aligned with the content. The final stage involves evaluating the generated questions based on a predefined set of criteria and selecting questions for classroom assessments. We demonstrated the proposed framework with a case study in which we generated different question formats using a Grade 6 science topic. We concluded by discussing the promises and limitations of generative AI for generating classroom assessments.