AI-Driven Software Engineering: A Systematic Review of Machine Learning’s Impact and Future Directions

This study looks into the expanding importance of machine learning (ML) in software engineering, to provide a thorough evaluation of its applications, categorise existing approaches, and propose prospective areas of future research. As machine learning (ML) continues transforming software development processes, understanding its potential and limitations is critical for success. A systematic literature review was carried out utilising major scientific databases, including IEEE Xplore, ACM Digital Library, Scopus, Arxiv, and Web of Science. After applying strict criteria for inclusion and exclusion to an initial pool of 105 publications, 57 were chosen for further review. The study synthesised concepts from the reviewed literature using both quantitative and qualitative approaches, including thematic coding and statistical analysis of publishing trends. The findings highlight major applications of machine learning in software engineering, including code generation, error detection, and program maintenance. Furthermore, the paper identifies a growing trend in the usage of graph neural networks (GNNs) to analyse code architectures, which are validated by experimental evidence. These achievements demonstrate ML's transformational potential for optimising the software development life cycle. While machine learning offers significant opportunities for automation and optimisation in software engineering, challenges such as low model interpretability, high computation costs, and limited integration into existing workflows remain. Addressing these challenges is important to fully realise ML's potential. Integrating machine learning into traditional programming methodologies, utilising federated learning for privacy-preserving collaboration, and developing interpretable ML models targeted to software engineering roles are all intriguing research avenues.