Sarcouncil Journal of Engineering and Computer Sciences

Abstract: Modern supply chains are characterized by volatility and complexity, exposing the inadequacy of traditional management practices. Despite the promise of using Artificial Intelligence (AI)/Machine Learning (ML) in Supply Chain Management (SCM), the adoption is hindered by black-box decision models, integration challenges, and lack of interpretability. This paper provides a review of the current state of AI adoption in SCM, taking both academic literature, regulatory environments and corporate practices into account. Results indicate that AI has helped in advanced forecasting, automation, and operations efficiency; its benefits are constrained by explainability gaps, scalability issues, and lack of the right alignment to governance needs. Industry use-cases in the form of Amazon, Walmart, DHL, among others, demonstrate the practical nature of the use of AI/ML in SCM and the various bottlenecks of these approaches. This research proposes solutions to address these challenges, which consist of advancing forecasting using advanced ML techniques, leveraging on reinforcement learning (RL) and multi-agent RL to cope with adaptability, graph neural networks (GNNs) for resilience, infusing explainability and human oversight to drive compliance, and integrating sustainability measures to improve SC optimization. The paper highlights that a successful adoption of these solutions, with the required level of governance and organizational readiness, can ensure the supply chains resilience, transparency, and sustainable networks that can better operate under uncertainty and disruption.