Economic Evaluation of Artificial Intelligence-Driven Software Testing in Modern IT Enterprises

This dissertation provides a comprehensive empirical evaluation of AI-powered testing frameworks as a strategic intervention to simultaneously enhance software quality and reduce operational costs in continuous integration and continuous deployment (CI/CD) environments. Employing a mixed-methods quasi-experimental design across twelve enterprise software projects over six two-week sprints, the study compares conventional automated testing (Selenium/TestNG) against three AI-driven frameworks: self-healing test automation, predictive defect analytics using XGBoost, and intelligent test case generation. Quantitative metrics include defect escape rate, mean time to detect failure, flaky test rate, test maintenance hours, cloud execution costs, and failure analysis effort. Qualitative insights derive from thirty-five semi-structured engineer interviews. Results demonstrate a 42.7% reduction in post-release defect escape rate (p < 0.001), a 59.6% decrease in mean time to detect failure, a 78.6% reduction in flaky test incidence, and a 58.7% decrease in total cost of ownership for quality assurance—translating to annualized savings of $1.45 million across the twelve projects. Self-healing automation contributed 48% of total cost savings, while predictive defect analytics accounted for 35% of quality improvements. Implementation challenges include initial model training requirements of three to four weeks, necessary upskill in feature engineering, and occasional visual misidentification of legitimate UI changes. The findings conclusively establish AI-powered testing as a high-return investment, with break-even achieved within two to three sprints for organizations maintaining more than two hundred regression tests. The study concludes with a phased adoption framework, actionable guidelines for QA managers, DevOps engineers, and executives, and directions for future research including large language model-based test generation and longitudinal organizational impact studies.