Generative artificial intelligence for construction: Use cases, trends, challenges, and opportunities

Document Type

Article

Publication Date

10-15-2025

Abstract

Recently, generative artificial intelligence (AI) technologies such as Generative Adversarial Networks (GANs), large language models (LLMs), Generative Pre-trained Transformers (GPT), and diffusion models have been increasingly applied to address challenges and inefficiencies within the Architecture, Engineering, Construction (AEC) workflows, particularly in design, planning, and construction management. However, its current research landscape remains fragmented, with limited synthesis of trends and unclear pathways for adoption in the construction industry. To address these gaps, a mixed-method review is conducted, combining bibliometric analysis to quantitatively map research trends with qualitative thematic synthesis for in-depth contextual insights. A total of 148 publications were retrieved from Scopus and Google Scholar (2014–2024). The bibliometric analysis identified 49 high-frequency keywords, grouped into six thematic clusters, characterizing the quantitative research landscape. Complementing this, the qualitative synthesis examined five dominant application domains: (1) proactive safety monitoring and risk prevention, (2) generative AI for sustainable construction, (3) automating design through generative intelligence, (4) construction education, and (5) construction management, within which key research gaps and practical challenges are critically examined. Building upon these insights, the study proposes four-level research roadmap spanning (1) industry-level considerations, (2) organizational and stakeholder perspectives, (3) project-level perspectives, and (4) technological integration. Unlike prior reviews that concentrated on isolated single-model technologies or narrowly defined domains, this study offers a comprehensive, cross-domain analysis of generative AI for construction. By employing a mixed-method review—integrating quantitative bibliometric mapping and qualitative thematic synthesis—it bridges technical, organizational, and implementation perspectives to deliver a holistic understanding of the field. Hence, this review offers clear avenues for future investigation, empowering researchers to expand and refine Generative AI toward achieving a more efficient, resilient, and sustainable construction.

Publication Title

Journal of Building Engineering

Link to Full Text

Share

COinS