A MacArthur “genius” unearthed the secret images that AI uses to make sense of us
Turns out there’s a bit of Caravaggio in artificial intelligence. Also a little Vermeer, Francis Bacon, Dalí…you could even see something of Edward Hopper, if you’re looking for it. A fascinating new series of images reveals what computers “see” after being fed images and symbols from Western literature, philosophy, and history, and they’re tantalizingly familiar....MUCH MORE
Though they may look it, these weird, vivid scenes aren’t the visions of Old Masters. They’re the hallucinations of our future masters—the product of artificial intelligence algorithms taught to create art by (human) artist Trevor Paglen, who won a $500,000 MacArthur “genius” award for his work just last week.
Artist, geographer, now genius
Even geniuses work hard. To hone his artistic practice after earning his MFA, Paglen went back to school and got a PhD in geography from the University of California-Berkeley. He has learned scuba diving in order to photograph the undersea cables that carry data from continent to continent. He has developed new photography techniques to catch surveillance planes and satellites passing high overhead, and has risked arrest to photograph secret military bases whose very existence is denied by governments around the world.
The 43-year-old’s self-directed mission is to spotlight the quiet structures of data-exchange, surveillance, and automation all around us. In his latest attempt to turn his camera on the world’s watchers, Paglen goes deep inside the automated brain, to reveal what computers see when they’re watching us, and what they dream when they’re alone. “What does the inside of the cloud look like?” Paglen said to me, describing his new show at Metro Pictures in New York City. “What are artificial intelligence systems actually seeing when they see the world?”
The installation, called A Study of Invisible Images, displays pictures that were never intended for human eyes: Archival photos used by researchers to train real computer-vision algorithms; annotated images that show how an AI processes landscapes, faces, or gestures; and the curiously baroque art generated by AI itself.
The Adams and Eves of computer vision
Paglen describes AI-training images as the “Adams and Eves” of computer vision. These include portraits created from the US military’s Facial Recognition Technology (FERET) program originally compiled in the 1990s to teach computers to recognize faces and foundational to the development of computer vision since then. I didn’t recognize any of them and perhaps that’s the point—shouldn’t the ordinary concerned citizen, Paglen seems to be asking, have a little fluency with the visual vocabulary of some of the most powerful automated technologies in the world?...
On Saturday it was "The Case of the Mafia and the Stolen Caravaggio". And previously:
Finance and Art: "When the Caravaggio No One Thinks Is a Caravaggio Becomes the Basis of an Asset Swap"
Questions America Is Asking: "Is there a €120m Caravaggio in your roof?"
Art: "Caravaggio and the Experts: Science v. Connoisseurs"
Banking: Ducats for Caravaggio, The Genoa Connection and Medici Moolah
Quite a few money - tech - Caravaggio connections actually. One that still cracks me up:
Caravaggio, The Cardsharps, c. 1594.
"Twas ever thus.