Artificial Intelligence-Enabled Test Automation for Reliable and Efficient Software Quality Engineering

Abstract

The increasing complexity, scale, and dynamic behavior of contemporary software systems have intensified the limitations of conventional rule-based testing approaches. Artificial intelligence (AI), particularly machine learning, deep learning, reinforcement learning, and intelligent decision-making techniques, provides a basis for developing adaptive test automation systems capable of selecting actions, prioritizing test scenarios, learning from execution outcomes, and responding to changing system conditions. This research and review paper examines how AI-enabled decision-making principles can be conceptually transferred to automated software quality engineering. The analysis is based exclusively on the supplied literature, which primarily investigates intelligent decision-making, reinforcement learning, deep reinforcement learning, autonomous maneuvering, trajectory planning, and adaptive control in unmanned aerial vehicle environments. Although these studies are not directly concerned with software testing, their methodological foundations provide useful analogies for intelligent test selection, adaptive execution, test prioritization, and autonomous quality decision-making. The paper develops a conceptual AI-enabled software testing framework consisting of software-state representation, intelligent test generation, reinforcement-based test selection, adaptive execution, defect-oriented prioritization, and continuous feedback. The findings indicate that reinforcement-learning-based decision mechanisms are particularly relevant for environments where testing decisions must be repeatedly optimized under changing conditions. However, the transfer of these techniques requires careful treatment of differences between physical autonomous systems and software environments. The paper concludes that AI-enabled test automation can improve testing efficiency and adaptability when learning mechanisms are combined with controlled validation, risk-based decision criteria, and human oversight.