The UN built an AI tool that maps disaster damage from space to speed relief to the hardest-hit places

The United Nations Development Programme has built an artificial-intelligence tool, called the Rapid Digital Assessment or RAPIDA, designed to answer a brutal question in the hours after a disaster: where is the damage worst, and where should help go first? When an earthquake, flood or storm strikes, responders often lack the basic information needed to direct aid, and reaching remote areas to survey them in person can take days that survivors do not have.

RAPIDA tackles that gap by combining several data sources into a fast picture of the situation. It draws on geographic information systems, high-resolution satellite imagery, social-media signals and even night-light data, and uses AI to interpret them, in part by feeding satellite images into a land-use model to detect how a landscape has changed. By layering in information about building materials and the dimensions of structures, the system can estimate the number of damaged homes and the tonnage of debris that will need clearing.

“When an earthquake, flood or storm strikes, responders often lack the basic information needed to direct aid, and reaching remote areas to survey them in person can take days that survivors do not have.”

The tool has already been used in a real emergency. After the Herat earthquake in Afghanistan, RAPIDA helped responders narrow down which locations were most affected so that scarce in-person assessment teams could be sent where they mattered most, saving time and gathering data on hard-to-reach, remote areas. UNDP has paired such tools with other AI systems, including a deployment platform that matches skilled personnel to needs on the ground based on expertise, availability, proximity and languages.

The caveats are important. A remote damage estimate is a starting point for response, not a substitute for verification on the ground, and satellite-and-social-media analysis can be thrown off by clouds, sparse data or noisy online signals. AI here supports human responders and local authorities; it does not make the decisions or deliver the aid. But giving relief teams a faster, clearer map of where the worst harm has fallen, especially in remote places that are easy to overlook, is a concrete way for technology to help people in their most vulnerable moments.