Answers are grounded in this paper and cite their sources.
The integration of artificial intelligence (AI) into software testing has emerged as a key mechanism for optimizing code quality and reducing quality assurance costs; however, the proliferation of algorithms and the heterogeneity of their applications hinder the informed selection of techniques and the consolidation of a unified body of knowledge. This study presents a systematic literature review, conducted in accordance with the Kitchenham and Charters guidelines and the PRISMA statement, analyzing 93 primary studies published between 2020 and 2025, retrieved from four academic databases (Scopus, IEEE Xplore, Web of Science, and ScienceDirect) and filtered through ten inclusion and exclusion criteria alongside an 80% quality threshold. The results show that supervised learning-based defect prediction and test case generation assisted by Transformers and large language models (LLMs) represent the two most mature research streams in this corpus, where hybrid and ensemble methods report the highest precision and coverage indicators within their respective domains. Systematic gaps are also identified in non-functional testing, test oracle evaluation, reinforcement learning beyond test case prioritization, and explainable AI applied to software testing. The resulting taxonomy provides a reference framework for quality assurance teams and delineates a priority research agenda for the field.
No takes yet. Share an insight, caveat, or question.
Avila et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: