Enhancing Question-Based Learning with AI : Generating and Evaluating Formative Questions and Feedback in Programming Education

Sammanfattning

This master thesis explores the potential of Artificial Intelligence (AI), specifically ChatGPT, to enhance learning in programming courses through Pure Question Based Learning (pQBL). The research was conducted in an introductory programming course at KTH. A Skill Map was developed using ChatGPT to generate formative questions and feedback across three alternative question types: Single Output Questions (SOQ), Ranking or Ordering Questions (ROQ), and Fill-in-the-Blank Questions (FBQ). Data was collected through the OLI Torus Learning Platform and student surveys. The generated questions were mapped to specific skills, and the feedback provided was detailed and process-focused. The results showed that ChatGPT could effectively generate formative questions and feedback, enhancing student engagement and understanding. The study underlines the necessity of generating a considerable amount of AI-generated questions to meet teachers’ standards. Notably, the results highlight that FBQ questions required minimal editing, while the other question types needed more substantial revisions. Student surveys indicated positive attitudes towards the AI-generated content and the pQBL approach. This study implies that ChatGPT shows promise as a tool for generating formative questions and feedback in programming courses, potentially reducing the need for extensive human labor and lowering educational costs.