Democratizing human-AI collaborative decision-making in agri-food supply chains: a trust-building framework

The black-box nature of Artificial Intelligence (AI) hinders stakeholder trust and comprehension, limiting practical adoption in business contexts. This study draws on findings from a case study in the agri-food sector applying Explainable AI (XAI) tools to support decision-making in case of SC disruptions. It proposes a trust-building framework that integrates ensemble learning models, interpretation tools (e.g., SHapley Additive exPlanations (SHAP), Large Language Models), interactive visualizations, and dynamic feedback mechanisms. These elements enhance transparency, enabling users to comprehend and act on AI-generated insights tailored to their expertise levels. By emphasizing human-AI collaboration, the framework addresses key gaps in accessibility and usability, empowering diverse stakeholders to engage with and benefit from XAI systems.