The Impact of Artificial Intelligence on Supply Chain Resilience: A Study of AI-Based Demand Forecasting and Inventory Optimization

Global supply chains have been subjected to an unprecedented sequence of shocks over the past several years, from the COVID-19 pandemic to geopolitical conflict, trade policy uncertainty, and extreme-weather events, exposing the fragility of lean, just-in-time operating models and elevating supply chain resilience to a board-level priority. Artificial intelligence (AI) has emerged as one of the most widely discussed enablers of resilience, with machine learning, deep learning, and, more recently, generative AI and large language models being deployed across demand forecasting, inventory optimization, risk detection, and logistics coordination. This paper presents a structured literature review of the relationship between AI adoption and supply chain resilience, with particular emphasis on two of its most mature application domains: AI-based demand forecasting and AI-based inventory optimization. Drawing on approximately fifty peer-reviewed articles, conference papers, and industry reports published primarily between 2020 and 2026, the review synthesizes evidence on how AI technologies improve forecast accuracy, reduce stockouts and overstocking, and support the visibility, agility, and viability capabilities associated with resilient supply chains, including during the COVID-19 disruption. The review also consolidates the benefits reported in the literature—cost reduction, error reduction, and faster decision cycles—alongside recurring challenges, including data quality and integration, the black-box or low-explainability nature of many AI models, high implementation costs, workforce resistance, and a persistent research gap concerning the long-term, human-centric, and multi-tier integration of AI within resilient supply chain ecosystems. Building on these findings, the paper proposes a conceptual framework linking AI-enabled forecasting and inventory capabilities to supply chain resilience outcomes, and concludes with managerial implications and directions for future empirical research.