Machine learning-driven processes in architectural building design

Machine learning (ML) has emerged as a transformative technology in the construction industry, enhancing the performance of predictive systems to facilitate earlier decision-making. This paper reviews 230 papers selected from 2706 articles to examine ML applications in architectural building design (ABD) processes, providing both bibliometric and qualitative analyses. Qualitative analysis examines the design domains, purposes, ML methods, and design stages of applied ML applications, revealing a growing focus on building performance and autonomous design generation. To support this, an extended ML-ABD workflow is proposed, integrating insights from state-of-the-art ML applications and addressing advances in generative ML systems. This offers guidance to construction stakeholders on the challenges and opportunities within the design processes, supporting the shift towards more intelligent and innovative workflows. The paper serves as a foundational resource for advancing ML-driven methodologies to enhance construction design processes, in which generative ML shows potential for automating workflows in the ABD process.