[2503.13310] Generative AI for Software Architecture. Applications, Challenges, and Future Directions

View a PDF file from the paper entitled AI Tawlidi for software engineering. Applications, challenges and future trends, written by Matteo Esposito, Siyahu Lee, Sergio Maurishini, Nanuman Ahmed, Thomas Serney, Cratik Vidianathan, Valentina Lenardi and David Tipi

PDF HTML (experimental) view

a summary:Context: Genai artificial intelligence (GENAI) transforms a lot of software development, however its application in the structure of software is still in its cradle, and no previous study systematically dealt with this topic. Objective: We aim to systematically synthesize use, logical basis, contexts, ease of use and future challenges of Genai in software structure. Method: We conducted a review of multi -foci literature (MLR), analyzing the literature that peers and gray reviewed, identifying current practices, models, adoption contexts, reported challenges, and extracting topics through open coding. Results: Our review has set a great adoption for Genai to support architectural decision and architectural reconstruction. Openai GPT models are often applied, and there is a fixed use of technologies such as a few generation of shots and retreats (RAG). Genai was mostly applied to the initial stages of the SDLC’s life development cycle (SDLC), such as structures and architecture to the symbol. Homogeneous structure and microservice were dominant goals. However, the precise test of Genai’s outputs was usually missing from studies. Among the most frequent challenges are typical accuracy, hallucinations, moral aspects, privacy issues, lack of data collections for architecture, and the absence of sound evaluation frameworks. Conclusions: Genai shows great potential in software design, but there are still many challenges on their way to increase adoption. Research efforts should target general evaluation methodologies, deal with morals and accuracy, increase transparency and ability to clarify, and enhance data groups and standards for architecture to fill the gap between theoretical possibilities and practical use.