Applicability of artificial intelligence for improving infrastructure resilience.

Abstract

Infrastructure systems are susceptible to extreme weather events, climate change, ageing, interdependence and cyber threats, highlighting the inadequacy of conventional resilience approaches. This study aims to review and provide an integrated conceptual perspective on how artificial intelligence (AI) capabilities contribute to infrastructure resilience capacities and resilience lifecycle stages. To examine the relationship between AI capabilities and resilience characteristics, a qualitative narrative literature review was conducted following a thematic analysis. The review proposes an integrated conceptual framework linking AI technologies, AI-enabled capabilities, 4R resilience capacities, and resilience lifecycle stages. The findings imply that AI facilitates capabilities such as hazard prediction, real-time monitoring, damage detection, predictive maintenance, and optimised decision-making, supporting the core resilience capacities (4R: robustness, redundancy, resourcefulness, rapidity) to achieve different resilience lifecycle milestones, including prevention, absorption, recovery and adaptation. The findings also suggest that AI has the potential to bring a transition from conventional reactive, siloed approaches towards a more proactive, adaptive and data-driven resilience system. The findings further reveal the importance of AI integration in infrastructure planning, management, and disaster risk governance to enhance preparedness, monitoring, adaptation and decision-making. However, there are several constraints which limit AI performance in resilience, including data quality and availability, governance issues, cybersecurity risks, system interdependencies and complexity, and a lack of interdisciplinary integration. Future research must focus on ageing infrastructure systems, multi-sector interdependent infrastructure, and explainable AI approaches to enhance transparency and understanding of AI-enabled decision-making.