AI in Software Development Research

Overview

The problem

AI tools like GitHub Copilot and ChatGPT are widely used in software development, but most research focuses on productivity gains. Less attention goes to developer behavior, long-term skills, over-reliance, reduced code comprehension, and security and ethical risks, and there is little clear guidance for responsible AI use in real projects.

What I built

• Wrote a research proposal examining how AI-assisted tools affect code quality, developer productivity, and the challenges of adopting them.
• Reviewed 21 peer-reviewed studies (2019–2025) from academic journals and conferences, using a systematic review and qualitative thematic analysis.
• Grouped findings into key themes: developer productivity, code quality, bug and error detection, and developer acceptance and challenges.
• Identified a research gap in human–AI collaboration, developer learning and skills, long-term dependency, and ethical AI use.
• Planned a 15-week timeline across planning, refinement, analysis and writing, and finalization phases, with a budget of RM125.

Result

The review found that AI tools bring clear benefits, with significant productivity and quality improvements, but that over-reliance and reduced code understanding remain real challenges. It concludes that balanced and responsible AI use is essential, and it offers guidance for students, developers, and organizations.