Alp Sungu

Using AI to improve the environmental and nutritional efficacy of food subsidies

Food consumption is a key lever for enhancing human health and environmental sustainability on earth. Our goal in this research is to better understand and find effective approaches to jointly alleviate malnutrition and climate-change outcomes in developing countries. In-kind food subsidies are a primary policy intervention in economically underprivileged communities. However, despite the proven efficacy of machine learning and data analytics in various industries, these technologies have not been widely adopted to enhance the design and impact of such subsidies. We aim to bridge this gap: this study develops a method to design nutrition- and climate-targeted food programs i.e., we algorithmically decide which foods to subsidize to maximize the nutrition value and minimize the carbon footprint of a subsidy program. This project builds on extensive field data collection in an underserved community in India, which comprises a randomized controlled trial, collection of large-scale point-of-sale transaction data, and survey data. Our methodology will combine a predictive choice model based on machine learning techniques and an optimization algorithm. Through this research, we aim to offer policymakers a tangible, data-driven pathway to refine food subsidy schemes, achieving healthier populations and a reduced carbon footprint.

The undergraduate RA is expected to have strong proficiency in coding, and basic understanding of statistics and/or econometrics.