AUTOMATED SOFTWARE TESTING USING ARTIFICIAL INTELLIGENCE: TECHNIQUES AND FUTURE PERSPECTIVES

Software testing is an essential phase of software development that helps identify defects, verify functionality, and improve software reliability. However, traditional testing approaches often require considerable time, manual effort, and maintenance, particularly in complex applications. Artificial intelligence (AI) offers new opportunities to automate testing activities by learning from software data, generating test cases, predicting defects, and optimizing testing processes. This paper presents a narrative review of AI-based automated software testing techniques, including machine learning, deep learning, reinforcement learning, natural language processing, and large language models. It discusses their applications in test case generation, defect prediction, test case prioritization, regression testing, and test script maintenance. The paper also examines important challenges, including data quality, model reliability, explainability, computational cost, and integration with existing software development workflows. Finally, it identifies future research directions involving trustworthy AI, adaptive testing, standardized evaluation benchmarks, and human–AI collaboration. The review indicates that AI can support more efficient and adaptable testing, although human oversight and systematic validation remain necessary for dependable software quality assurance.