Artificial Intelligence supported Software Engineering: a comparative study of its application in the Software Development Lifecycle

Abstract

This thesis examines how Artificial Intelligence (AI) can be integrated into the Software Develop- ment Life Cycle (SDLC) to support both technical and managerial tasks. While AI tools are widely used for coding and testing, their broader role in requirements analysis, planning, documentation, and decision- making remains inconsistent and lacks clear, practical guidance. The work combines a theoretical review covering AI evolution, SDLC models, adoption patterns, and associated risks, with a comparative research methodology designed to evaluate modern AI systems such as GPT, Gemini, Claude, and DeepSeek. The experimental part uses a single case study and standardized prompts to evaluate each model on multiple SDLC artifacts. The evaluation is based on accuracy, completeness, consistency, interpretability, efficiency, governance compliance, and practical usefulness. This approach allows a structured comparison of how each AI model performs in different SDLC phases. The research primarily seeks to clarify how AI can boost efficiency, collaboration, and decision support in contemporary software engineering.

Această lucrare analizează modul în care Inteligența artificială poate fi integrată în Ciclul de Viață al Dezvoltării Software pentru a sprijini atât activitățile tehnice, cât și cele manageriale.