A Survey of Intelligent Programming and Code Optimization Techniques in Software Development

Abstract

Software development is continuously evolving with the integration of Artificial Intelligence (AI) techniques that enhance productivity, software quality, and development efficiency. This survey reviews intelligent programming and code optimization techniques applied across the Software Development Life Cycle (SDLC), including requirements analysis, coding, testing, deployment, and maintenance. It examines traditional software development approaches and modern architectures, followed by intelligent programming techniques based on Machine Learning (ML), Natural Language Processing (NLP), Deep Learning, Transformers, and Generative AI. The survey is limited to intelligent source code completion, AI techniques for code generation, classical optimization methods such as code cloning, invariant code motion, dead code elimination, local optimization, global optimization, and inter-procedural optimization. Recent studies demonstrate that AI-assisted approaches can improve code quality, development efficiency, vulnerability detection, energy efficiency, cost estimation, and software testing. However, challenges related to data dependency, interpretability, developer trust, technical integration, and the reliability of AI-generated code remain significant. The survey highlights current research trends and identifies future opportunities for developing reliable, explainable, scalable, and efficient AI-driven programming and code optimization solutions for modern software development