Answers are grounded in this paper and cite their sources.
Global supply chains are increasingly exposed to large-scale, interconnected disruptions, making supply chain resilience (SCRES) a critical strategic priority. Artificial intelligence (AI) is widely discussed as supporting risk anticipation, visibility, and adaptive decision-making. This paper presents a structured narrative synthesis of 75 studies that met the inclusion criteria and were published from 2015 to 2025. The search was conducted through January 2026 in Emerald, Web of Science, and Google Scholar. Stage-specific retrieval counts, search logs, Web of Science index names, the Google Scholar bounding rule, reference-manager software, the number of screeners, and an agreement statistic were not retained and cannot be reconstructed. The paper is therefore not presented as a fully auditable systematic review or as a PRISMA 2020-compliant review. Thematic coding is combined with a limited descriptive bibliographic summary of the same corpus. Citation networks and keyword-evolution maps were not produced. The synthesis suggests that AI is discussed as supporting preparedness, agility, and recovery primarily through predictive analytics, machine learning, digital twins, and AI-enabled decision support systems. Relative to published AI-SCRES reviews, the unique increment is a conversion architecture rather than another catalog of tools. Agility and adaptability are the capability paths through which AI is associated with resilience outcomes. Governance maturity and digital readiness constrain whether those paths can operate. Five bounded contributions follow from that architecture, including a capability-stage mapping, the separation of the two paths from the two constraints, a treatment of explainable AI as a decision-relevant requirement, and a specification of sensing, seizing, and transforming as subprocesses that future work can test. The practical implication is that AI is more credibly treated as a governed socio-technical capability than as a stand-alone algorithmic solution.
No takes yet. Share an insight, caveat, or question.
Aazami et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: