Download PDFOpen PDF in browserLarge Language Model Integration in Construction Safety: A Literature Review10 pages•Published: June 2, 2026AbstractThe construction industry continues to face persistently high rates of injuries and fatalities despite decades of safety initiatives, underscoring the need for innovative, data-driven approaches. Recent advances in artificial intelligence, particularly Large Language Models (LLMs), have opened new avenues for enhancing safety through automation, hazard recognition, and intelligent decision support. This study presents a comprehensive scientometric and thematic analysis of 60 peer-reviewed journal papers published between 2020 and 2025 that explore the intersection of LLMs and construction safety. The bibliometric results reveal a sharp growth in publications beginning in 2024, coinciding with the rise of generative AI technologies. Thematic analysis identifies five major research domains: text-based safety analytics, multimodal sensing and real-time monitoring, knowledge-enhanced reasoning, generative AI for data augmentation, and human-AI interaction for safety training and decision support. Key challenges include data scarcity, limited domain-specific reliability, and practical barriers to implementation. Overall, this study maps the intellectual landscape of LLM applications in construction safety, highlighting both current progress and future opportunities for leveraging language-based AI to achieve safer, more resilient construction environments.Keyphrases: ai, construction, construction safety, llms, multimodal safety system In: Wesley Collins, Anthony Perrenoud and John Posillico (editors). Proceedings of Associated Schools of Construction 62nd Annual International Conference, vol 7, pages 1202-1211.
|

