CIB Conferences
Abstract
Hazard recognition is a core capability in construction safety, yet conventional training often provides limited exposure to realistic, dynamic risk conditions. This paper presents an integrative review of 67 peer-reviewed studies on artificial intelligence-enabled hazard recognition training in construction. The review examines how immersive simulation, neuro-biometric sensing, mixed reality and AI-driven safety knowledge systems contribute to safety learning. Four research streams are identified: neuro-biometric analysis of hazard perception, adaptive immersive training, mixed-reality support for situational awareness and AI-based safety knowledge engineering. The synthesis indicates a shift from instructional safety training toward adaptive, data-informed learning ecosystems in which behavioural, physiological and operational data are used to personalise feedback and update training content. The paper also addresses implementation use cases and risks, including privacy, surveillance, cognitive overload and uneven organisational capacity. The review contributes a concise framework for understanding AI-enabled safety learning as a socio-technical capability rather than a stand-alone technological intervention.
Keywords
Construction safety; Hazard recognition; Artificial intelligence; Safety training; Immersive technologies
Recommended Citation
Ngo, Van-Yen; Nguyen, Quan T.; Nguyen, Bao-Ngoc; Le, Hoai-Nam; Luu, Quang-Phuong; Pham, Van-Hoan; Tran, Ngoc-Tuyen; and Aziz, Zeeshan
(2026)
"Artificial Intelligence For Hazard Recognition Training In Construction: An Integrative Review,"
CIB Conferences: Vol. 2
Article 61.
DOI: https://doi.org/10.7771/3067-4883.2219