Quantifying the Risks of Generative AI Integration in Construction Risk Management: A Fuzzy Logic Perspective

Abstract

This study investigates the socio-technical risks associated with integrating Generative Artificial Intelligence (GenAI) into Construction Risk Management (CRM). A three-stage methodology was adopted. First, a structured systematic literature review was conducted using Scopus to identify socio-technical risks through database and journal identification, keyword-based screening, and conventional content analysis. Second, a questionnaire survey was developed and administered to 136 construction management experts in the UK to evaluate the significance of the identified risks. Third, a fuzzy set theory-based risk quantification model was developed and implemented in MATLAB to handle uncertainty in expert judgment and to compute an overall fuzzy risk number for ranking. Twelve socio-technical risks were analysed, representing human capability, adoption dynamics, and governance and security concerns. The results indicate that insufficient training and data breaches are the highest priority risks, followed by a lack of awareness and unclear responsibility and accountability. The study provides actionable insights for practitioners and policymakers by supporting targeted mitigation strategies focused on workforce readiness, data governance, and accountable decision structures, thereby enabling responsible GenAI integration in CRM.