Generative AI in Programming Courses : Understanding University Student Usage and Learning Impact

As generative AI (genAI) tools like ChatGPT become increasingly integrated into educational contexts, understanding their role in programming education is necessary. While existing research focused largely on the technical performance of genAI tools, there is limited insight into how such tools affect learning outcomes in code-writing contexts. This thesis investigates how university students use genAI tools for code- writing in programming courses. Drawing on the Technology Acceptance Model (TAM), Self-Determination Theory (SDT), and Bloom’s Taxonomy, the research explored three main questions: how students use AI tools, what motivates them, and whether the AI enhance or hinder their ability to write and understand code. A mixed- method approach was used, including a controlled user study (N = 22), a post-study survey (N = 22), and semi-structured interviews (N = 10) with students at Jönköping University. The findings showed that programming experience and motivation play a key role in how students work with AI. Beginners more likely copied code without changes, while experienced students often modified and improved the code. Students driven by extrinsic motivation used AI more passively, while those with intrinsic motivation engaged with it more actively. While AI tools can make coding tasks easier, they may also reduce deeper learning if not used thoughtfully. Based on the results, a guideline with recommendations for educators on how to support students in using AI tools in ways that build programming skills and independent thinking was developed.