Exploring Ethical Awareness and Practices in Software Development : Study of AI Coding Assistant Tools Usage Among Developers
Wijaya, Alvin Steven (2025)
2025
Master's Programme in Computing Sciences and Electrical Engineering
Informaatioteknologian ja viestinnän tiedekunta - Faculty of Information Technology and Communication Sciences
Hyväksymispäivämäärä
2025-12-15
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-2025113011082
https://urn.fi/URN:NBN:fi:tuni-2025113011082
Tiivistelmä
Artificial intelligence (AI) coding software such as GitHub Copilot, Codex, and to an extent ChatGPT are rapidly changing the way humans code software with real-time code suggestions to facilitate automating coding, speed up productivity, enhance creativity, and optimize efficiency. However, their adoption creates pressing ethics, law, and security challenges. This thesis examines how programmers perceive and address these issues with a focus on ethical consciousness, trust, accountability, and governance in using AI-based coding software.
The study employs a qualitative exploratory design, supported by a pilot questionnaire and seven semi-structured interviews of developers representing a range of different roles and areas. Overall trends in adoption and subtle reflection on ethics are both available through this approach. The findings offer a "trust paradox": developers continue to utilize AI assistants despite recognizing repeated errors, security exposures, and intellectual property issues. While they valued such technologies for accelerating repetitive work and prototyping, they had limited faith in production-level or high-stakes application. Ethical awareness was highest for privacy, intellectual property, and data protection but lowest for bias and fairness. Responsibility was perceived as distributed: developers thought of themselves as primary ethical agents but also emphasized organizational policy and vendor controls as essential to risk mitigation.
The research identifies disconnects between high-level ethical framework and everyday development practice. The developers are operating without formal organizational control, rather employing individual judgment with ad hoc controls such as code review, static analysis, and test environments. The study argues that closing the gaps will mean multi-level governance strategy that combines human-in-the-loop review, organizational policy, and vendor disclosure. Lastly, this thesis contributes to current discussions on the ethical integration of AI in software development by providing empirical data on the attitudes and practices of practitioners regarding how they enact ethical responsibility in their day-to-day work.
The study employs a qualitative exploratory design, supported by a pilot questionnaire and seven semi-structured interviews of developers representing a range of different roles and areas. Overall trends in adoption and subtle reflection on ethics are both available through this approach. The findings offer a "trust paradox": developers continue to utilize AI assistants despite recognizing repeated errors, security exposures, and intellectual property issues. While they valued such technologies for accelerating repetitive work and prototyping, they had limited faith in production-level or high-stakes application. Ethical awareness was highest for privacy, intellectual property, and data protection but lowest for bias and fairness. Responsibility was perceived as distributed: developers thought of themselves as primary ethical agents but also emphasized organizational policy and vendor controls as essential to risk mitigation.
The research identifies disconnects between high-level ethical framework and everyday development practice. The developers are operating without formal organizational control, rather employing individual judgment with ad hoc controls such as code review, static analysis, and test environments. The study argues that closing the gaps will mean multi-level governance strategy that combines human-in-the-loop review, organizational policy, and vendor disclosure. Lastly, this thesis contributes to current discussions on the ethical integration of AI in software development by providing empirical data on the attitudes and practices of practitioners regarding how they enact ethical responsibility in their day-to-day work.