Generative artificial intelligence and mathematical modelling for eco-friendly humanitarian supply chains

Abstract

Humanitarian supply chains (HSCs) are critical for delivering relief items to affected communities in disaster response. One challenge in HSCs is the environmental impact of the waste generated by the packages of relief products. A promising approach to addressing this challenge is to use circular approaches in HSCs, aiming to reuse, repurpose, and recycle packages, thereby minimising waste in target areas. However, optimising the material flow in HSCs with circular considerations is not trivial given the uncertainties and complexity of the response. This study presents a decision support system for humanitarian field-based decision-makers to balance environmental impact with effective, efficient, and equitable distribution of relief items. The system integrates generative artificial intelligence (Gen-AI) with an evidence-based multi-objective mathematical model that supports determining material flows during the response. We apply our methodology to a real dataset from the humanitarian response to the 2022 Pakistan floods, collected from various textual reports published by involved humanitarian organisations across different outlets, such as their websites, common humanitarian databases, and blog posts. Our findings indicate that combining Gen-AI and mathematical modelling is a scalable approach that enables the incorporation of circularity insights from extensive textual data to reduce the environmental impact of humanitarian response.