Application of artificial intelligence for identification and automation of software defect correction

DOI: 10.31673/2412-9070.2026.017403

Abstract

In the light of global technological growth trends and increasing complexity and scale of software, the challenges of maintaining and optimizing processes for existing software are becoming increasingly pressing. This article examines current methods and approaches of effective applying of artificial intelligence (AI) for the identification and automation of the processes of analyzing and correcting software defects and failures. Significant attention is given to the analysis of modern machine learning models and generative algorithms, which enable effective error detection in program code, assessment of their severity, and automated correction of identified issues with minimal involvement of developers and engineers. The appropriateness of using AI in real-world projects is substantiated, especially in the context of large-scale software development and maintenance projects, where factors such as accuracy, operational speed, and cost reduction for testing and support are taken into account. In particular, the potential practical application of generative realizations of large language models (LLM), such as ChatGPT, is analyzed for automating the analysis and subsequent correction of software defects. As a result of the research, practical approach to effective integration of AI into the workflows of software development is proposed, the prospects for its implementation in the industry are identified, and the main challenges associated with the use of such technologies are highlyghted. The practical results of the accomplished research confirm the effectiveness of using AI to improve software quality, reduce the impact of the human factor, and enhance the productivity of development teams.

Keywords: artificial intelligence; software development optimization; software support optimization; defect correction; crash and error analysis; machine learning; generative AI models; large language models.