Comparing Open Source with Software Code Generated by AI Tools from Software Maintainability Quality Factor Perspective

Abstract

Artificial Intelligence (AI) has made great strides in various industries, including software development, with tools such as ChatGPT transforming the way code is written, maintained, and optimized. This study examines the impact of AI-generated code on software quality, with a focus on maintainability, code complexity, and documentation quality. Comparing AI-generated code with open-source code from GitHub for three tasks of varying difficulty (easy, medium, and hard), we evaluated key metrics, including the maintainability index (MI), lines of code (LOC), cyclic complexity (CC), Halstead volume (V), and comment ratio. The findings indicate that AI-generated code is usually more verbose its cyclical complexity tends to drop on easier tasks, reducing error rates. In a complex task maintainability prefers to support programmers with AI-generated code significantly, and better documentation according to comments. These results show that AI tools can support and enhance code quality, especially, in an industry where maintainability and simplicity are critical.