Communicating with a Programming AI Assistant Through Voice : How Voice Interaction Could Enhance the User Experience for Programmers

Sammanfattning

The rise of AI programming assistants, such as GitHub’s Copilot, has transformed how programmers interact with code, making it more accessible to less experienced users. This thesis explores how voice interaction can shape the programming experience through the development of Astra, a locally hosted, voice-interactive AI assistant. Which was created to support research through design methodology and evaluation with novice programmers. Usability testing revealed that voice interaction can enhance flow, collaboration, and creative exploration for some users, while others preferred the control of manual input. This culminated with the removal of the visible code interface from participants, which encouraged more conversational and exploratory communication. This thesis's findings suggesting that future AI programming assistants should adapt dynamically to user preferences and balance creative exploration with technical accuracy. With the study contributing to interaction design discourse by showcasing how voice interaction can integrate into programming workflows and identifying opportunities for future development.