Breaking the Black Box by Exploring Technical and Ethical Challenges in AI-Driven Software Defect Detection
Rajeev Ranjan Kumar, Kripa Shanker, Rakshit Ranjan · Advances in computational intelligence and robotics book series · 2025
AI is challenging the idea of how software defects are detected as before it has created a way to generate new ways of how software defects are detected and predicted by organizing test cases, detecting anomalies of the code and getting test and ideas of the bugs in a manner that in the past never existed. Yet, as well as these developments may go hand in hand with a trade-off a majority of the AI models are so-called black boxes, having little degrees of transparency in the mechanisms of decision-making. This chapter explores technical complexities as well as ethical issues that are witnessed when using AI in software quality assurance. We discuss various AI practices applied in defect detection, namely supervised learning, deep learning, and generative models and also issues like data quality, model explainable qualities, and integration. We explored such ethical issues as bias, responsibility, and the influence of society on the role of developers. We conclude that we address emergent solutions based on explainable AI (XAI), regulation and human-AI collaboration strategies.