Averting Crises

Every day the condition of millions of acres of farmland is captured by satellite and aerial imagery, picking up data invisible to the human eye but indispensable for predicting crop yields, monitoring farmland health, and forecasting shortfalls in food production. At the same time, historical satellite data is being linked to known conditions on the ground, helping predict the consequences of disruption.

Together, this information allows stakeholder—technical experts, regional or local decisionmarkers, crsis-affected communities, and food producers—to more effectively predict, adapt to, and mitigate the impacts of impeding crises.

Hunger is on the rise in many places around the world, driven most severely by conflict and exterme weather. Survey and field data are critical for models to accurately predict hunger under such conditions. These data can benchmark and validate new AI-enabled tools to ensure that outputs match realities on the ground. As reductions in funding for humanitarian programming constrain data collection efforts, AI's ability to faacilitate—but not replace—primary data collection will be critical.

The more training an AI-enabled model has using relevant localized data, the better it performs. There is no substitute for this primary data.

Food insecurity results from complex interactions between social, economic, environmental, and technological drivers. Modeling these dyanmics accurately requires the analysis of a wide range of data types over long time spans and diverse geographies. Early warning systems such as headlines, market signals, and boiological markers of population health are broadening the scope of information models rely on.

Through the collection and integration of discrete information, AI-enabled data collection. andmodeling tools. arenow helping identify impending shocks and escalating risks well before they occur.