Leveraging Generative AI Models to Improve Software Engineering Productivity: A Comparative Study of OpenAI's Codex, Google's Gemini, and China's DeepSeek

Manuja Sanjay Bandal

Abstract: The rapid advancements in Generative AI (GenAI) are reshaping software engineering by streamlining code generation, improving software quality, and reducing development cycles. Among the leading AI models in this domain, OpenAI?s Codex, Google?s Gemini, and China?s DeepSeek each bring distinct advantages to software development. This paper presents a comparative analysis of these models, evaluating their effectiveness in automating coding tasks, debugging, documentation, and optimization. Furthermore, we propose an integration framework to incorporate GenAI into the software development lifecycle (SDLC) and conduct empirical assessments to measure its impact. Our findings indicate that AI-driven development enhances efficiency by accelerating coding processes, improving software maintainability, and reducing errors. However, concerns such as security vulnerabilities, long-term maintainability, and region-specific AI regulations pose challenges that must be addressed. This study concludes by highlighting key areas for future research in AI-assisted software engineering.

Keywords: Generative AI, Software Engineering, Code Automation, AI-assisted Development, Software Productivity, DeepSeek, OpenAI Codex, Google Gemini