AI-Assisted Model For Generating Multiple-Choice Questions

Tetiana Krushynska, University of Jyväskylä

Jani Ursin, University of Jyväskylä

Ville Heilala, University of Jyväskylä

Date

02/2026

AI Overview

Why it matters

Writing multiple-choice questions that evaluate knowledge application in real-world situations is challenging and time-consuming for educators, especially in natural science education.

Study at a glance

Question: How can teachers and AI collaborate to generate quality multiple-choice question prototypes and transform them into series?
Setting and data: Simulations using university-level biology textbooks and exam learning objectives, May to August 2025, generating 352 test items.
Method: Simulation study evaluating prompt engineering workflows against quality criteria (goal-oriented conceptual modelling).
AI tools: MS Copilot (GPT-4), ChatGPT (GPT-4.1), DeepAI (GPT-4), Wrizzle (GPT-4o mini)

Key findings

Approximately half of the generated multiple-choice question prototypes and series items were acceptable without editing based on nine strict quality criteria.
Distributing a three-step generation process across multiple AI models produced higher quality question prototypes than using a single model alone.
Minor post-generation corrections moderately increased the acceptance rate of question series and raised prototype acceptance to 70% to 85%.

Summary drafted with AI from the full paper and may contain errors. See the full paper for details.

Multiple-choice questions (MCQs) are widely used across diverse educational fields and levels. Well-designed MCQs should evaluate knowledge application in real-world situations. However, writing such test items in sufficient numbers is challenging and time-consuming, especially in natural science education. The problem of a sufficient number of MCQs has two aspects: content coverage and exam security. Therefore, generating test items involves two tasks: creating MCQ prototypes and transforming these prototypes into item series. In automated item generation, prototype creation aligns with template-based methods like cognitive modelling, while item expansion corresponds to example-based techniques. The aim of this research was designing the goal-oriented conceptual model of human - AI co-creation of MCQs that should meet strictly formulated quality criteria. The resulting three-step model for creating MCQ prototypes distributed prompts between several AIs, with human revision of responses for each prompt before setting the next one. To transform the MCQ prototype into an MCQ series, a one-step model was developed in which multiple new items are generated simultaneously. These items assess the same learning outcome but are not simple rephrasings of the prototype or of one another. Based on human and automated evaluation, approximately half of the output MCQs were acceptable without editing. Minor corrections of initially rejected test items allowed for a moderate increase in acceptance of MCQs in series and a significant improvement of MCQ-prototypes.