Artificial intelligence (AI) has significant potential to enhance demand forecasting accuracy, responsiveness and resilience in supply chains. Yet adoption remains uneven due to systemic and interdependent barriers. This study investigates the hierarchical structure and cascading influence of technological, organizational and environmental barriers to AI-enabled forecasting.
A cross-sectional survey of 162 supply chain professionals across multiple industries was conducted to evaluate fourteen barriers identified from prior literature. Data were analysed using interpretive structural modelling (ISM) to establish a hierarchical structure of interdependencies and Matrice d’Impacts Croisés Multiplication Appliquée à un Classement (MICMAC) analysis to classify barriers into driving, linkage, and dependent categories. Exploratory factor analysis (principal axis factoring with varimax rotation) was employed to empirically validate barrier clusters within the technology–organization–environment framework.
The results reveal a three-tier cascade of barriers. Infrastructure gaps, real-time data deficits, skill shortages, cost constraints and integration complexity emerged as root-level drivers. Algorithmic opacity, ethical concerns, vendor dependency and scalability limitations function as volatile linkage barriers, while resistance to change, rigid processes and awareness deficits were found to be dependent outcomes. These findings highlight that behavioural resistance is structurally induced rather than autonomous, and interventions must prioritize root causes before cultural or strategic concerns can be effectively addressed.
The ISM–MICMAC framework provides a diagnostic and prioritization tool for managers and policymakers. For practitioners, the results emphasize sequenced adoption roadmaps: investing first in infrastructure, skills and financial planning; stabilizing linkages through transparency, explainability and governance mechanisms, and only then targeting behavioural adaptation. For policymakers, the findings point to systemic interventions such as investment in digital infrastructure, incentives for small and medium-sized enterprises, skill development programmes and explainability standards for operational AI.
This study advances AI adoption research by moving beyond static barrier checklists to model hierarchical interdependencies in forecasting contexts. It extends the TOE framework through integration with ecosystems theory and disruptive innovation, offering a mid-range theoretical explanation of how structural barriers cascade into organizational and behavioural consequences. The research contributes both an empirically validated barrier architecture and an actionable roadmap for overcoming adoption challenges in supply chain forecasting.
