In the recent Spike Jonze film “Her,” a lonely man buys a futuristic Siri-style computer program designed to interact fluently with humans. The program’s female voice speaks intimately to its owner through an earpiece, while also organizing his e-mails and keeping track of his appointments, until eventually he falls in love with it. Watching it happen is unsettling because of how well the program works on our human protagonist—how powerfully this computerized, disembodied simulation of a woman affects him.
The piece of software at the heart of “Her” only exists within a work of science fiction, of course—one set in a comfortably vague point in the future. While today’s smartphones and computers do “talk” to their users, they do so without the emotionally potent and slightly unpredictable qualities that might make a machine feel human.
But in certain quarters, automated beings with that exact set of qualities have already started to emerge. For the past few years, tiny computer programs have been telling jokes, writing poems, complaining, commenting on the news, and awkwardly flirting—on the Web and through the medium of Twitter, where their messages live side by side with those composed by the carbon-based life-forms who take daily delight in their antics.
One of the most prolific makers of these little programs—or bots, as they’re known—is 30-year-old Darius Kazemi, a computer programmer who lives in Somerville and works at a technology company called Bocoup, while devoting himself to the veritable petting zoo of autonomous digital creatures he has invented in his free time.
Chances are you haven’t heard of Kazemi. But over the past two years, he has emerged as one of the most closely watched and pioneering figures at the intersection of technology, cultural commentary, and what feels to many like a new kind of Web-native art.
Kazemi’s Twitter bots—as well as the interactive websites, games, and simple computer programs that make up their extended family—run constantly, entertaining and provoking their fans by riffing on the algorithms that increasingly run our world. One project, Professor Jocular, finds comments people have made on Twitter and tries, clumsily, to explain what’s funny about them, even when nothing is. Another generates a steady stream of awkward pickup lines. Another pulls from a public database of “last words” uttered by condemned prisoners in Texas, and, to heartbreaking effect, displays one by one every sentence in which an inmate on the precipice of death used the word “love.” Kazemi’s crowning achievement so far might be the program he rigged up to order $50 worth of completely random books, CDs, and DVDs from Amazon and deliver them to his house every month. Over Christmas, he sold a limited run of “prints” of this project, allowing his fans to receive their own randomly selected gifts.
Kazemi’s dozens of projects have won him admirers among a range of people so wide it suggests the world doesn’t quite have a category for him yet. His work is tracked by video game designers, comedians, philosophers, and fellow bot-makers; an English literature professor named Leonardo Flores has written about the output of his bots in an online journal devoted to electronic poetry. Web designer Andrew Simone, who follows Kazemi’s work, calls him “a deeply subversive, bot-making John Cage.”
Kazemi is part of a small but vibrant group of programmers who, in addition to making clever Web toys, have dedicated themselves to shining a spotlight on the algorithms and data streams that are nowadays humming all around us, and using them to mount a sharp social critique of how people use the Internet—and how the Internet uses them back.
By imitating humans in ways both poignant and disorienting, Kazemi’s bots focus our attention on the power and the limits of automated technology, as well as reminding us of our own tendency to speak and act in ways that are essentially robotic. While they’re more conceptual art than activism, the bots Kazemi is creating are acts of provocation—ones that ask whether, as computers get better at thinking like us and shaping our behavior, they can also be rewired to spring us free.
When most of us think about artificial intelligence, we tend to imagine entities that humans can interact with—like HAL from “2001: A Space Odyssey,” or the chess-playing computer Deep Blue. But the truth is that in the modern world, “intelligent” algorithms are constantly running behind the scenes in ways that affect what we see, what we do, and even where we go. Google tracks our interests in order to shape our search results; Amazon’s software tells us what products we might want to buy next. In new skyscrapers, elevator dispatch systems automatically group people who are going to similar floors, thereby determining who we see every day.
“There’s just all sorts of automated beings all around us, and they don’t necessarily have to pretend like they’re human, either,” Kazemi says. “Our traffic light system—that’s a giant automated system that we’re constantly navigating and responding to.”
Kazemi was in high school in Fairfax, Va., when he first attempted to create a program that interacted with—meaning, actually talked to—people in the real world. He was in an engineering class that met every Friday to talk shop over pizza. He bought a voice synthesizer, a special chip that could read words from a computer out loud, and designed a program that would call Domino’s at a particular time every Friday and place the order without any human input whatsoever. Except the 15-year-old Kazemi chickened out before sending his robot into the world. “In the end, I didn’t have the guts to plug it into an outlet,” he says.
After graduating from the Worcester Polytechnic Institute, Kazemi entered the world of video game development, building programs that could systematically test new games for bugs. Kazemi also designed his own games—like many game designers, he considered games an art form as much as a technical accomplishment—until one day in 2012, he decided that the medium was holding him back from what he really wanted to express. It was around this time that Kazemi read a book of philosophy called “Alien Phenomenology, or What It’s Like to Be a Thing” by Ian Bogost, a professor of interactive computing at the Georgia Institute of Technology. In the book, Bogost advanced a concept that greatly appealed to Kazemi: that it was possible to be a philosopher who didn’t write down ideas, but instead made objects that embodied them.
The “objects” Kazemi was moved to make after reading Bogost’s book were Twitter bots, a class of digital beings typically associated with irritating spam accounts that automatically send advertising messages to any Twitter user who mentions a particular word or brand name. Kazemi was hardly the first person to realize the potential in programming conceptually interesting Twitter bots—for example, Adam Parrish had already made the popular @everyword, which has been working its way alphabetically through the English language, tweeting one word every 30 minutes, since 2007. But Kazemi quickly became one of the medium’s most inventive practitioners.
Kazemi’s first foray into the field was called Metaphor-a-Minute. The way it worked was simple: The bot would pull nouns and adjectives from an online dictionary called Wordnik, and arrange them in a particular order so that each tweet presented a metaphor both bizarre and fleetingly plausible. (Examples: “a premonition is a warren: defenseless and tacit,” “an impression is a mucus: nondomestic, rootlike.”) The effect was that of a very smart but helplessly confused alien being trying to make sense of the English language. To date, the account has generated nearly half a million metaphors.
From there Kazemi was off to the races. He made a RapBot that used a rhyming database to write hip-hop verses. He created Amirite, a hammy jerk of a bot that makes corny, often nonsensical “am I right” jokes that sometimes strike a nerve: “Wendy Davis? More like Trendy Davis, amirite?” His Startup Generator lampooned tech culture with a constant stream of dubious business ideas (“Paypal for dropouts”). More recently he created his most popular bot to date, Two Headlines, which crawls the latest news stories on Google, picks two at random, and switches important keywords to generate a series of broken windows into the popular conversation: “Beirut seeks love advice from Katy Perry”; “Iran Is Working On Smart Contact Lenses That Can Monitor Your Body’s Health.” Bogost now considers himself part of Kazemi’s growing fan base, waiting for the next bot to be born. “You have a favorite comedian or favorite artist and you look forward to what they say, because you want to see the world through their eyes,” Bogost said. “The same kind of thing is happening with Darius.”
At a coffee shop near his office, Kazemi says he feels about his bots the way he imagines parents must feel about their children. “I’ve created these things, and they’re kind of separate from me now, and so I do feel kind of proud of them,” he says. “Every morning I wake up and I look at the last two hours of TwoHeadlines, and it just gets me every time.”
In the brief time since the birth of the form, bots have spawned their own subculture, with a blurry line between creators and fans. Rob Dubbin, a bot-maker who is also a writer for “The Colbert Report,” created a website called The New York Review of Bots; Kazemi recently hosted a Bot Summit where people from all over the world discussed such things as the ethics of bots and bot taxonomy.
If this sounds a bit esoteric, just think of “Her,” and remember how much public anxiety there is right now regarding the future of human-robot relations. For Dubbin, that anxiety is part of what makes Kazemi’s art so effective: He seizes on what makes most people uncomfortable about robots—that they’re unpredictable, that they are prone to misinterpret their instructions and end up doing unthinkable things—and somehow uses it to make them delightful and easy to relate to. “He turns the thing about algorithms that’s unsettling into something that’s beautiful and surprising and funny and insightful,” Dubbin says. Paraphrasing a remark made by another bot-maker, Brett O’Connor, he added: “His bots are not speaking from emotion, but they can say things that trigger emotional reactions. And that is maybe a little uncanny.”
Kazemi usually goes into his projects thinking of them as just funny experiments—that was how he first conceived of Amazon Random Shopper, for instance. But after receiving his first shipment—a copy of a Noam Chomsky book on linguistics and a CD by a Hungarian avant-garde composer he’d never heard of—and giving a few interviews on the subject, he started to think of it in other terms, too: as a critique. Kazemi had created a way to declare independence from Amazon’s recommendation engine. By designing a bot capable of ordering a completely random thing, Kazemi was breaking through the “filter bubble” that increasingly shapes his—and everyone’s—experiences.
The special power that Kazemi brings to this mission is what he and others call “procedural literacy”: Like most computer programmers, he can see past the surface of the digital tools in our lives, and figure out the automated mechanisms by which they’re making decisions on our behalf. (“Darius kind of sees the matrix a little bit,” Dubbin says.) And while he resists being cast as some kind of activist, Kazemi does see his work as a way to expose the mechanics of automation by putting it front and center in surprising new ways—“sort of an attempt to educate people,” he says, “and prompt them to do some of that thinking on their own.”
That spirit runs through many of his projects. In that sense, they fall squarely in the tradition of so-called generative art—think Ellsworth Kelly’s random color paintings or the cut-and-paste experiments of William Burroughs. But Kazemi’s work stands apart by virtue of its immersion in the blinking and chaotic landscape of the Internet. This immersion in the Web is crucial, because the way people use it—to communicate, to shop, to get their news—is so often the target of his satire. His programs riff relentlessly on clichés we recognize; one even builds entire Web pages that look eerily like genuine human-built Buzzfeed listicles, as if to say, “A computer could do this.” As Andrea Shubert, a video game designer and longtime acquaintance of Kazemi’s put it, “He has a knack for knowing where to poke, and creates interesting ways to poke.”
It is also possible to see his work as nostalgic for a younger, less predictable Internet. Mark Sample, a fellow bot-maker and a digital studies professor at Davidson College, said in an e-mail that much of what Kazemi makes is aimed at recapturing the serendipitous spirit of the Web before browsing became quite so slick, curated, and targeted. “I see Darius’s work as an ongoing critique of this contemporary Internet, a reminder that the Internet doesn’t have to be this way.”
In this light, Kazemi and his bot-making friends can be seen as exploring a medium through which we now do much of our everyday business—and then rerouting the wiring that underlies that medium, in a way that moves us to question how we normally use it. By making works that don’t just take advantage of Internet technology, but use it to reveal the invisible rules of the Web, Kazemi may have found nothing less than a new kind of public art for the 21st century—changing, self-referential, and in its insistent randomness, oddly alive.
Leon Neyfakh is the staff writer for Ideas. E-mail firstname.lastname@example.org.