Generative AI in Software Architecture and Design: A New Paradigm for Intelligent Systems Engineering

Main Article Content

Surendra B Konathala

Abstract

This paper explores the transformative potential of Generative AI in software architecture and design. By integrating insights from research in program synthesis, architectural search, and architectural decision automation, we demonstrate how GenAI can significantly enhance the accuracy, efficiency, and scalability of software development processes. Through a comparative study, we evaluate experimental GenAI models against traditional search-based and heuristicbased design approaches, focusing on key indicators such as design quality, time-to-solution, and optimization. Our findings, illustrated through program fragments and prototypical applications in areas like program synthesis and neural architecture generation, suggest that GenAI introduces a paradigm shift, facilitating not only the construction of robust architectures but also the rapid generation and optimization of diverse design options. Integrating established software engineering principles such as modularity, cohesion, and low coupling into generative models further enhances the maintainability and scalability of the resulting architectures.

Article Details

Section
Articles