How I Became the Face of Facial Recognition
Much of the stock photography depicting the tech world is terrible, or at least highly inaccurate and staged. An infamous stock photo shoot from 2016, for example, shows models of various genders and racial backgrounds using a soldering iron to repair a circuit board.
The photos look great. Except the models are holding the iron by its element, not its handle. The element of a soldering iron gets to about 300 to 800 degrees Fahrenheit when the iron is in use. If the photos were real, the soldering iron would have scalded the models. They’re also soldering the wrong side of the board. From the photos, it’s clear that no one involved in the shoot had ever used a soldering iron, much less repaired a circuit board.
At least someone behind those photos tried to portray a tech concept accurately. Too often, photographers depict complex concepts like hacking or coding by getting a young, white, male model, dressing them in a hoodie, putting them in a darkened room, and shooting creepy photos of them hunched over a screen filled with green binary numbers.
While some coders do memorize hexadecimal sequences, very few people have coded using binary numbers since the punch card era. And the stereotype of hackers and coders as pimply, hunched-over, white male loners is inaccurate. Most hackers and coders today are probably suit-wearing career professionals operating out of a drab, government, or corporate-funded office park, not prepubescent teenagers in a basement.
Tech is already mysterious to most people. And with that mystery comes apprehension or anger. It’s easy to read all of the world’s evils onto Facebook or Twitter if you’ve never personally grappled with the existential challenges of moderating user-generated content or running an advertising-supported business. By portraying techies using creepy stereotypes and abstractions, photographers risk inadvertently worsening the public’s perception of tech as abstract and unknowable and techies as warped and inhuman.
In many cases, actual tech is way more accessible — and way weirder — than many non-techies think. A common method for soldering certain types of circuit board components — even in relatively high-tech labs — is to lay out your board and then stick the whole thing on an electric skillet, like the kind people use to make pancakes. Some labs use a toaster oven instead. Photos of a PCB roasting on a skillet might not scream “high-tech.” But it’s a much more accurate portrayal of what electronics repair is really like than a model misusing a $10 soldering iron the photographer picked up at Home Depot.
With good research — or ideally hands-on experience — accurately photographing technological objects is at least comparatively easy. Physical tech objects like micromobility scooters, for example, are ubiquitous, colorful, and lend themselves to visual metaphors — like when they’re dumped on a sidewalk or thrown through a window. But how do you depict a technology that’s often deliberately obscured and shadowy, like facial recognition? These technologies are invisible. Yet they’re powerful and insidious enough to determine who gets a loan, gets a job, or gets arrested.
Because technologies like facial recognition are more a process or a feature than an actual object, they’re hard to photograph. In 2019, though, I decided to try. I took a selfie while walking my dog, and ran it through Google Vision, an artificial intelligence API from Google which includes much-used facial recognition features. The system layered a series of bounding boxes and dots around my face, highlighting my facial features, the outline of my head, and more. This is standard stuff in the facial recognition world. I published the image on several stock photography marketplaces.
Since the image went online in late 2019, it’s been used in more than 50 publications and networks all over the world, including Yahoo, Vox, Der Spiegel, CBS, the New York Times, the New York Observer, Inside Higher Ed, Forbes, Business Insider, Digital Trends, and even OneZero right here on Medium. Quite inadvertently, I’ve become the face of facial recognition.
Recognizing the fact that powerful, dangerous new technologies might look visually boring is vital for today’s tech consumers.
The results have been strange. Family members routinely reach out to say that they’ve seen me in a news story. When I research a new facial recognition technology for a story here on Medium, I often click through on a Google News result only to find my own face staring back at me. Ironically, the photo is one of the few of me that Clearview AI didn’t find and add to my profile.
The surprising reach of my facial recognition image says several things about the industry — and about the underlying tech itself.
Firstly, there’s a real hunger for photos of hard-to-depict technologies. Journalists spend a lot of time covering technologies like facial recognition. Yet too often, we’re forced to use overly blunt symbols (a closeup of a surveillance camera, some flashing police lights on a dark street) or total abstractions (an illustration of a face breaking apart into lots of tiny pieces) to illustrate these technologies, rather than photos which actually show the technologies themselves. Accurate photos of complex technologies provide a lot of value.
My photo also captures the strange fact that technologies like facial recognition are simultaneously totally earth-shattering and extremely boring. Facial recognition can facilitate human rights atrocities or the mass violation of civil liberties. Yet technologically, the whole field basically amounts to drawing a bunch of dots on a person’s face and then measuring the distance between those dots. For a world-altering technology, it’s much less visually interesting than you might expect.
My facial recognition photo isn’t even especially good. It was shot on a cellphone. I look creepy. The background is a collection of random cars and my neighbors’ trash cans. But at least it’s honest. Facial recognition doesn’t look like a suave, slightly bearded man staring at a super-thin cellphone while it scans his face with a laser, creating a 3D version that smiles back at him. It looks like a bad selfie with a bunch of green dots on it. I think there’s a power in showing that — just as there would be power in showing hackers as besuited office workers or PCB repair on a pancake skillet.
If you’re a photographer, think about better, more accurate ways to depict today’s technologies. It’s a tough challenge and one I’m still working on myself. And if you’re a consumer of tech content, remember to think critically whenever you see a new technology depicted visually. If you read lots of news articles that show facial recognition as a flashy, space-age process involving robots and lasers, you might assume that it’s a futuristic technology that is years away.
It’s not. Technologies like facial recognition are here now, and they’re already being used to track and profile you. Recognizing the fact that powerful, dangerous new technologies might look visually boring is vital for today’s tech consumers. It’s a reminder that you won’t necessarily know these technologies when you see them. You have to remain vigilant in other ways — politically, legally, and culturally — if you want to spot and combat them.
It’s also vital to remember that tech people don’t necessarily look the way you might expect. Depictions of hackers and coders as young male loners are not only inaccurate, but they also send the implicit message that women, older adults, the socially connected, and non-whites couldn’t possibly be members of a technologically advanced community. That’s a stereotype we absolutely have to fight — visually and otherwise.
When photographers show models holding a soldering iron wrong, it’s a funny, mostly harmless goof. But other depictions of tech have the power to dehumanize and obscure — or to clarify and enlighten.