Sponsored by: Computer Science Department Intended Audience(s): Public Categories: Lectures & Seminars Abstract: As Covid-19 spreads in low and middle-income countries, economic disruptions have left hundreds of millions without work or income, precipitating the first rise in global extreme poverty in over 20 years. To offset the pandemic’s most devastating effects, national policymakers and humanitarian organizations are scrambling to provide emergency humanitarian aid to those who need it most. But determining “those who need it most” is difficult in many poor and conflict-affected countries, where official government registries are often incomplete and out of date. This talk describes ongoing work that leverages recent advances in machine learning, applied to rich data from satellites and mobile phone networks, to target and deliver emergency social assistance. The algorithms we have developed now form the basis for Covid-19 response programs in Togo and Nigeria, which are providing subsistence cash transfers to hundreds of thousands of poor families. More broadly, this work illustrates the important role that machine learning can play in the future of humanitarian response. It also highlights several open technical challenges – as well as important social and ethical considerations – that arise at the intersection of machine learning and applied economics.