It’s no surprise that Amazon and the CIA are interested in your data, but the new information they’re looking to collect may seem a bit off-brand. DigitalGlobe, alongside chipmaker Nvidia, is partnering with Amazon and the venture arm of the CIA to map out the constantly evolving parts of urban landscapes — including trash heaps.
Using SpaceNet, an algorithm based on the same technology that fuels the facial-recognition software used by companies like Facebook, DigitalGlobe will eventually map high-resolution images of up to half a million square miles of the Earth’s surface. There’s one big difference, though: Instead of collecting information about buildings, like Google Maps, SpaceNet will train its A.I. to algorithmically collect precise data on objects such as waste and commercial traffic.
The MIT Review of Technology reported recently that DigitalGlobe released its first set of data, a detailed set of images of the Rio de Janeiro that depict the city in up to 50-centimeter resolution. The images released only outlined the city’s buildings, but as the SpaceNet A.I. grows smarter it will be able to identify more distinct data. Such information could be used to improve public services —like trash collection — or even track economic output in regions, MIT reported.
This isn’t the first time the CIA has partnered with Amazon. In 2014, the CIA gave Amazon Web Services $600 million to develop a computing cloud that services all of the agencies in U.S. intelligence.
According to its website, Digital Globe’s services include providing data to humanitarian relief efforts as well as offering “defense and intelligence” surveillance.
So what interest does the CIA have in making your garbage collection more efficient? It’s hard to say. The project is being supported by CosmiQ Works, the division of the CIA’s venture arm which focuses on space (aka satellite spying).
AWS now represents over half of Amazon’s earnings, so striking a deal with the highest-funded government intelligence agency only boosts their value. Amazon constantly seeks to one-up its competitors and this project could easily exceed Google Earth as the most comprehensive global geographic data set.