Technology acceptance of GitHub Copilot: A study among computing students

Abstract

AI pair programmers (AIPPs) are becoming increasingly used in software development environments. This is encouraging universities to examine the role of these tools in computing education and programming courses to ensure that their programs remain market-oriented. However, despite the rapid adoption of AI pair programmers, limited empirical evidence exists regarding students' acceptance and educational use of these tools in authentic project-based software development environments. One of the most widely used AIPPs is GitHub due to its integration with many development environments. This study investigates the Technology Acceptance (TA) of GitHub Copilot among computing students. We conducted a mixed-methods study with 114 senior computer science and software engineering students from three public universities in the Kurdistan Region of Iraq. The task was to develop a static-responsive portfolio website using GitHub Copilot as an AI pair programmer. The Technology Acceptance Model (TAM) was used to evaluate Perceived Ease of Use (PEU), Time Efficiency (TE), Support in Learning (SiL), Code Quality (CQ), and Willingness to Recommend (WtR). Quantitative descriptive statistics and thematic analysis of qualitative responses were used to evaluate these features. Findings indicate high positive initial acceptance. Thematic analysis showed that students value GitHub Copilot for overcoming syntax barriers and reducing development time, but also express concerns about understanding and over-reliance. This study presents one of the first empirical mixed-methods examinations of AIPP acceptance in computing education. The contribution of the study lies in identifying the dual nature of this acceptance: high enthusiasm tempered by reflective concerns about deep learning. The results suggest that Copilot can improve efficiency in programming courses, but its use should be balanced to avoid superficial dependency on the tool.