Has Google finally grown up? The care with which it has handled facial-recognition technology seems to support this thesis. Compare it with Facebook. When Zuckerberg’s social network unveiled its facial-recognition technology in June, it found itself in the middle of a global privacy backlash. But Google has avoided that fate: A few weeks ago, it unveiled a technology to automatically identify one’s friends in photos uploaded to Google+—and almost nobody noticed.
The different reactions are easy to explain: Facebook enabled this feature for all users without asking their permission, while Google made its tool optional. Facebook may now be warming up to this more-polite approach, too: Its recent settlement with the Federal Trade Commission stipulates that all future changes to existing privacy controls would require user consent.
The Web seems to be moving away from the “opt-out” mentality of the arrogant bully—e.g., “We know you’ll love this feature, so we’ll enable it by default!”—to the “opt-in” mentality of the smooth-talking diplomat—“Hey, check out this new feature—but only if you want.” As Facebook’s embrace of “frictionless sharing” shows, it’s one thing to force us to share by altering our privacy settings—and it’s quite another to convince us that sharing is something we really want to do. The former is an offense; the latter is a cause for celebration.
And yet this triumph of the “opt-in” is not all that it seems. While it’s certainly less coercive, any opt-in still makes the underlying technology—automated facial recognition, in this case—seem normal and acceptable. But no technology companies will acknowledge this. “The decision is all in the user’s hands.” “It’s all about giving users more control.” “We are not forcing anyone—people can stay out.” Such bland rhetoric of “user empowerment” has been the staple of Silicon Valley gospel for decades. It rests on a naive belief that technologies are just tools and their impact is quite narrow and limited to accomplishing (or not) the task at hand. Thus, if users want to use Tool X to accomplish Task Y, the only thing up for debate is the desirability of Task Y. That the wide adoption of Tool X may also trigger an unexpected Effect Z never bothers the instrumentalists or, if it does, they just write it off as something incalculable.
Alas, such reasoning overlooks the fact that technologies, in addition to serving their immediate functions, also have an ecological footprint—in that they can transform environments, ideologies, users, power relations, and even other technologies. While cars may be a perfectly effective way of getting from Point A to Point B, one shouldn’t focus on this feature alone and disregard what the car culture in general might be doing to the quality and even forms of urban living or pollution rates or mortality statistics. Focusing on the immediate uses of an artifact—regardless of whether those are “opt-in” or “opt-out”—seems like a poor way of navigating the “car problem.”
Similarly, to assume that a given technology isn’t problematic because its users can turn it off seems misguided. Why disregard the possibility that, once enough people opt in to use it, the collective adoption of this technology might dramatically transform the social environment, making nonuse difficult or impossible? Once enough Californians have opted in to use the car, something changed—both at the levels of public infrastructure and norms—that makes much of California completely inhospitable to carless living. The car still gets us from Point A to Point B, but wouldn’t our quality of life be much higher if we tried to anticipate its side effects by developing a more multifaceted view of the car technology?
Now, to return to the subject of automated facial-recognition technologies, here is what we know. This technology can be easily abused; a search engine that generates people’s names from their faces would be very popular with dictators, all too keen to crack down on popular protest. We also know that facial-recognition technology has already penetrated many walks of life. It is a popular way to secure our smartphones and laptops. It’s used in many game consoles to create a more personalized gaming experience. It’s used to track (and in real-time!) the number of male and female patrons in bars. And the list goes on.
Such seemingly innocuous uses beget a generation of start-ups that are looking for new uses for this technology—not all of them innocuous but many of them foreseen by its critics. By the time the general public wakes up, of course, this technology becomes so deeply embedded in our culture that it is too late to do anything.
In a sense, we are dealing with a process that is more sinister than the popular notion of the “butterfly effect”—the idea that the flap of a butterfly’s wings in Brazil can set off a tornado in Texas. Call it the “Palo Alto effect”: A carefree user in Palo Alto, Calif., who decides to “opt in” and use Google’s facial-recognition technology ends up strengthening a dictator in Damascus. Why “sinister”? Because the Palo Alto user, unlike the butterfly, can actually think two steps ahead—but prefers not to.
What’s to be done? Well, we can put the ethical burden squarely on Internet users and sensitize them to the ultimate (even if indirect) consequences of their choices. There are many precedents for this. Mounting concerns over economic inequality, climate change, and child labor have led to the emergence of the “ethical consumption” movement, which seeks to get consumers to consider the ethical ramifications of their behavior in the marketplace.
In a similar vein, why not think about applying similar concepts to our engagement with the Internet? What would “ethical browsing” or “ethical social networking” entail? Never using sites that exploit facial-recognition technology? Refusing to do business with Internet companies that cooperate with the National Security Agency? These are the choices we’ll have to make if we don’t want the Internet to become an ethics-free zone. After all, unreflective use of technology—just like unreflective shopping—does not a good citizen make.
But let’s not allow Internet companies off the hook, either. Of course, Google and Facebook are different from rapacious corporations exploiting poor farmers or underage children. Neither company is building surveillance tools that would be used by dictators. What they do, however, is help create the apposite technical and ideological infrastructure for such tools to emerge in a seemingly natural manner. This doesn’t provide strong grounds for regulation—but it opens the door for citizen activism, boycotts, and, if all else fails, civil disobedience.
Internet companies know perfectly well that they’ve got responsibilities. Earlier this year, Eric Schmidt, Google’s executive chairman, called facial-recognition technology “creepy” and expressed his concern about it. And yet Google has just endorsed this technology—albeit with the “opt-in” proviso. This, Google thinks, shields it from any accusations of unethical behavior; after all, it’s all up to the user! But would we be persuaded by oil companies claiming that anyone concerned with climate change doesn’t have to drive a Humvee? Perhaps not. It’s in pretending that they don’t know how this sad movie ends that technology companies’ chief ethical blunder becomes evident.