Generative AI in Programming Education (Dagstuhl Seminar 25311)

Abstract

This report documents the program and the outcomes of Dagstuhl Seminar 25311 "Generative AI in Programming Education". During the seminar, we examined the transformative impact of Generative AI on programming education. Because they can solve many introductory tasks given only natural language prompts, AI tools are challenging established approaches to programming education, in which there has been a traditional emphasis on writing small programs and providing (automated) feedback to learners. While these developments raise concerns about student over-reliance and inaccurate feedback, they also open opportunities for new pedagogical practices, such as fostering prompt literacy, adapting curricula, and designing AI-assisted learning tools. The present seminar convened 42 international experts to exchange knowledge, present research, and share innovations through keynotes, lightning talks, and tool demonstrations. Collaborative working groups explored implications for learning outcomes, assessment, equity, human values, and tool design, while identifying directions for systematic evaluation and interdisciplinary research. The seminar successfully established a foundation for a sustained community of practice and set an agenda for advancing programming education in the era of Generative AI.