With the goal of delivering critical information more quickly and cost-effectively, and with greater accuracy and detail, Atlas AI will use machine-learning algorithms and a variety of ground data to generate detailed estimates with respect to crop yields, economic activity, and poverty in sub-Saharan Africa. It will then validate its models against datasets collected by multilateral partners in the field, backed by peer-reviewed scientific research, and share that information with agencies and decision makers in a position to use it.
The effort is being launched in partnership with a team of Stanford University professors, including David Lobell, Stefano Ermon, and Marshall Burke, who had already shown that satellite imagery can be used to map poverty and crop yields. With support from the Rockefeller Foundation, Lobell, Ermon, and Burke began to build out the platform last year and consulted with partner organizations on the ground in Africa to test and operationalize new features, including high-resolution datasets on wealth, consumption, and agricultural yields.
"Atlas AI is an innovative model for translating the best research thinking into products and services that accelerate sustainable development," said Rockefeller Foundation vice president of innovation Zia Khan. "Part of the Rockefeller Foundation's vision is to unlock AI's tremendous potential to improve people’s well-being while mitigating downside risks."