An Exploratory Study of Student Use of Generative AI for Programming Based on a Vibe Coding Workshop

Abstract

Generative artificial intelligence tools are increasingly embedded in student programming workflows. Students use these tools to generate, explain, modify, and refine code through natural-language interaction. This development raises important questions about understanding, verification, responsibility, and the educational value of programming assignments. In this paper, we present an exploratory study of student perspectives on AI-assisted programming in the context of Vibe Coding for Engineers, a five-day workshop held at the University of Zagreb Faculty of Electrical Engineering and Computing. The study is based on anonymous pre-and post-workshop survey responses, with 75 responses in the initial survey and 52 responses in the final survey. The analysis examines how students use AI tools, how much code they expect AI to generate, how they check AI-generated code, how well they think they understand it, and how they view disclosure, responsibility, and acceptable use. The results suggest that AI-assisted programming was already common among the participants. Post-workshop respondents reported higher intended use of AI tools and expected AI to generate a larger share of their code. At the same time, they were also more likely to agree with the position that, if an AI-generated program works, they do not need to check in detail why it is correct. However, the post-workshop results also show an increase in students’ sense of responsibility for AI-generated code. The main message of the study is that students value AI tools because they make programming faster and easier. They also recognize that they must remain responsible for their work. However, in terms of learning and understanding, their position is less clear, they seem to balance pragmatic use of AI tools with the need to understand the code they submit.