Dear editors,
Technological advancement has opened new frontiers, bringing the logic of the market to bear on elements of the human experience that were hitherto considered off-limits. Artificial intelligence (AI) data processing tools promise a future where all data is made interpretable. Online platforms make money by observing and intervening in our behavior. The greater their ability to predict and modify it, the more easily they can sell their services as data brokers, advertisers, and surveillance specialists. Our digital footprints have long been conserved, but the very vastness of our data protected us for a time. Now that this data can be processed and made meaningful, our private lives have become exploitable and monetizable. Conservation has turned against us.
As Harvard Business School professor Shoshana Zuboff has argued, the turn towards algorithmic content, targeted advertisements, and the constant scraping of metadata muddy commonly held notions of the public and the private. Though mass culture has long shaped our identities, never before has it been so invasive. Online industries conserve our choices, experiences, and identities in the form of data, and this hidden preservation poses a threat to our private freedoms.
We now stand at the precipice of another technological sea change, one that could erode further the distinction between our innermost desires and data employed by advertisers. Much has been written about the way AI poses a threat to interpersonal relationships and about our fear of off-loading thinking to machines. But I am more concerned about AI’s ability to process data. Though many tech companies generate revenue by monetizing metadata, there is reason to believe that this resource is underexploited. Every day, we produce a near-endless stream of digital detritus. Algorithms parse through much of it, from the content of our searches to the amount of time our mouse hovers over a link prior to clicking it. Still, certain types of data, especially the content of non-text-based internet activity, remain much harder for algorithms to interpret. Algorithmic illegibility is one reason why Facebook, for instance, still relies on human beings to staff the content moderation teams that watch hours and hours of uploaded video in search of child pornography. Much of our illegible data is left inviolate.
AI gives us the ability to digest difficult-to-interpret forms of data. Tinder recently announced that it will give users the option to let AI algorithms scan their camera rolls and generate profile suggestions. Dozens of startups hawk AI tools they claim can watch and interpret online video footage. The company Flock Safety has brought this trend offline through the deployment, in cities across the United States, of their Condor Cameras—pan-tilt-zoom surveillance cameras that use AI to automatically focus on human passersby.
When viable, AI features like these will be deployed en masse to analyze data in new ways. The revenue model followed by tech companies demands it. I have no doubt, for instance, that models like the one developed by Tinder will peruse the photos we upload to Instagram or Facebook and intuit surprisingly personal assumptions from the most minute details.
If data can be fully interpreted, conservation becomes dangerous. Now that our private choices and digital identities are entirely accessible for exploitation, liberation might be freedom from conservation.
A skeptic may question the novelty of the coming change. It is true that there are already tools to track and use our data, and that the barrier separating the personal sphere from the rude logic that governs monetized behavioral intervention has already been pierced. Still, tech companies possess an incentive to acquire ever more intimate information about their users. With the use of AI tools, they will be able to make this data legible, and implement it into newly hypercustomized algorithms. We already know that Facebook uses metadata to assemble demographic profiles of its users, targeting ads based on ethnic identity. As a tool to expand this project, AI trained to interpret stubborn forms of data constitutes a new threat to our privacy. Our digital doubles will now approximate us as closely as the data allows. So please, stop feeding the fire. Conserve less.
Fleetingly,
William Moursund
Image: James McNeill Whistler, Wild West, Buffalo Bill, etching laid on paper, 1887 (National Gallery of Art, DC).