Artificial Intelligence and Architectural Design Before Generative AI: Artificial Intelligence Algorithmics Approaches 2000–2022 in Review

Abstract

This study explores the evolution of Artificial Intelligence (AI) applications in the AEC from 2000 to 2022, focusing on the transition leading up to the accessibility of Generative AI in 2023. Through a Bibliometric review of publications indexed in Scopus, Web of Science, and CumInCAD, the research examines the adoption of specific algorithmic approaches: Genetic Algorithms (GAs), Artificial Neural Networks (ANNs), and Agent-Based Systems(ABS). Findings reveal that GAs and ANNs exhibited comparable use until 2015, after which ANNs experienced exponential growth, surpassing GAs by 2016. ABS, although less prominent overall, saw a temporary surge starting in 2008, which established ABS as a distinct research category. Comparative analysis with CumInCAD highlights its early role as a primary repository for specialized research, surpassing WoS and Scopus untill quite a few years afterward. This research underscores key historical milestones marking AI's integration into the AEC, including advancements in evolutionary computation, machine learning, and distributed AI systems. While revealing critical trends, the study acknowledges its limitations, such as database bias and the exclusion of developments post-2023. Future research should extend beyond this period, incorporate qualitative analysis, and explore emerging tools in generative design to understand AI's growing impact.