If In late September Hurricane Ian battered Florida, causing destruction through high winds and flooding. A few weeks later, however, three counties hardest hit by Hurricane Ian saw some relief.
Nearly 3500 residents in Collier and Charlotte Counties got a push notification from their phones offering $700 cash aid, with no questions asked. An algorithm developed by Google and GiveDirectly based on satellite imagery had calculated that the people in need were living in severely damaged areas.
GiveDirectly is testing this new way of targeting emergency aid in collaboration with Google.org, the search and ad company’s charitable arm. These individuals were beneficiaries of Providers which manages food stamp payments. Targeting messages with help from AI software from Google allowed GiveDirectly to offer aid only to people who lived in areas devastated by Ian more quickly than manually sorting through the rolls of the app’s users.
Although this is the first instance of GiveDirectly using the technology in America, it was used in Togo for a similar purpose in the aftermath of the devastating pandemic that decimated the country’s economy. It offered assistance to poor households based upon signs of poverty, which were determined by image algorithms created at UC Berkeley and clues found in cell phones bills.
Four Google machine-learning specialists developed Delphi mapping software, which was used to power the Florida project. This tool was created by GiveDirectly and took six months to develop. This software overlays live storm damage maps with poverty data from the US Centers for Disease Control and Prevention to highlight communities that are in greatest need following disasters like hurricanes. Skai is another Google tool that provides storm damage information. It uses machine learning to analyse satellite imagery before and after disasters and to estimate damage to buildings.
“You now have a map that says where is socio-economically vulnerable, and where has been damaged,” says Alex Diaz, who leads Google.org’s AI for Social Good team. “That can help on-the-ground support and speed up delivery of aid.”
The algorithms that power Skai’s damage assessments are trained by manually labeling satellite images of a couple of hundred buildings in a disaster-struck area that are known to have been damaged. This software then can quickly detect all damaged buildings throughout the affected area. An academic workshop in 2020 on AI and disaster response presented a paper about the technology behind the software. The research paper claimed that the auto-generated damage assessment matches those performed by humans with 85 to 98 percent accuracy.
In Florida this month, GiveDirectly sent its push notification offering $700 to any user of the Providers app with a registered address in neighborhoods of Collier, Charlotte, and Lee Counties where Google’s AI system deemed more than 50 percent of buildings had been damaged. The offer has been accepted by 900 people so far. Half of them have received the money. If every recipient takes up GiveDirectly’s offer, the organization will pay out $2.4 million in direct financial aid.