An empirical analysis of GitHub Copilot

Sammanfattning: The rapid advancement of artificial intelligence (AI) has brought transformative changes to various fields, including software development. GitHub Copilot, an AI-powered code assistant developed by OpenAI in collaboration with Microsoft, has emerged as a promising tool that aims to enhance developers’ productivity by providing context-aware code suggestions. This bachelor’s thesis presents an empirical evaluation of GitHub Copilot, exploring its effectiveness and efficiency.  The study adopts a mixed-methods approach, combining both quantitative and qualitative techniques to assess GitHub Copilot’s performance in aiding software developers with solving work-related programming tasks. Additionally, after solving the tasks, a survey is conducted with the participants to gather their perceptions and overall impression of Copilot.  The results of the empirical evaluation provide evidence-based insights into the effects of using Copilot as a programming tool. Findings from the quantitative analysis reveal Copilot’s potential to significantly reduce development time, while also slightly increasing the code’s quality. Moreover, the qualitative analysis did not show any difference in the code’s efficiency when using Copilot.