P10: AI Generated Quizzes: AI-supported quiz generation for teachers
Translated title
P10: AI Generated Quizzes
Authors
Kristensen, Cecilie Städe ; Sørensen, Bob Mølgaard
Term
4. Term
Publication year
2026
Submitted on
2026-06-03
Pages
55
Abstract
This thesis presents a proof-of-concept system that creates multiple-choice quiz questions from PDF-based teaching materials using a local large language model (LLM), a type of AI that processes text. The system offers suggested questions as candidates that teachers can review, edit, and select before running the final classroom quiz. In a usability test with teachers, the workflow matched their use of quizzes as group-based, interactive activities. Feedback also showed that human review remains essential, especially when questions need changes in wording, relevance, or medical content. The thesis concludes that the system can support teachers in preparing quizzes from existing material, but it should be seen as a semi-automatic aid rather than a fully automatic quiz generator.
Afhandlingen præsenterer et proof-of-concept system, der genererer multiple-choice spørgsmål ud fra PDF-baseret undervisningsmateriale ved hjælp af en lokal stor sprogmodel (LLM), en type AI der kan bearbejde tekst. Systemet viser de foreslåede spørgsmål som kandidater, som læreren kan gennemgå, redigere og vælge, før den endelige quiz afvikles i klassen. En brugbarhedstest med lærere viste, at arbejdsgangen passede til deres måde at bruge quizzer på – som gruppebaserede og interaktive aktiviteter. Feedbacken viste også, at lærerens validering stadig er vigtig, især når spørgsmål kræver ændringer i formulering, relevans eller medicinsk indhold. Afhandlingen konkluderer, at systemet kan støtte lærere i at forberede quizzer ud fra eksisterende materiale, men bør forstås som et semi-automatisk hjælpemiddel frem for en fuldautomatisk quizgenerator.
[This apstract has been rewritten with the help of AI based on the project's original abstract]