This New Algorithm Writes Perfect “Artspeak”
Image courtesy of Selçuk Artut.
If you’ve ever read an artist statement or museum wall text hoping to develop a deeper understanding of the work, but come away more confused, you’re not the only one.
Istanbul-based artist Selçuk Artut has developed a tool to explore this familiar art world phenomenon. His machine-learning algorithm spews wall-ready artspeak at the press of a button. The code powers his latest artwork, Variable, in which a sculpture is accompanied by an automatically generated, wall-mounted electronic description à la art-world press release.
“There are all of these art pieces where people are trying to give a lot of meaning with the use of extensive texts,” rather than leave them open for interpretation, Artut tells me. And “there are plenty of examples of artists who are not coming up with clever ideas [in art] but who are really good at writing beautiful texts.”
Taking cues from websites like the empty-buzzword-spouting “Bullshit Generator,” Artut wondered if he could create something similar for the art world.
The project has its roots in 2013, when Artut was writing his Ph.D. dissertation on the philosophies of Martin Heidegger, and found himself struggling to get through the philosopher’s difficult 1927 magnum opus, Being and Time. (One Amazon reviewer describes the book as ideas “buried beneath an impenetrable barrier of incomprehensible jargon.”)
But it wasn’t until this past summer when Artut, who often uses coding as a tool for his artistic practice, took online courses on machine learning and machine intelligence from Stanford University and became inspired to apply the technology to his work. He would, he decided, “teach the machine to think like Heidegger.”
Photo by Murat Durusoy. Courtesy of Selçuk Artut.
To do so, Artut trained a computer with the text from Being and Time. The resulting algorithm formulates Heidegger’s words and ontological paradigms into three-sentence-long statements that sound all too similar to art world gibberish.
Variable consists of a wall-hung metal sheet behind which a network of eight small computers are hidden. Through corresponding holes, the computers project eight-letter words, like “movement” or “approval.” These become the artwork’s randomly generated “titles.” Nearby, a small box containing a five-inch LCD display is positioned where a wall label would normally appear. Viewers can press a button to see the title change and the screen animate with an entirely new digital artist statement.
When the work was displayed in a recent group exhibition at the Zorlu Performing Arts Center in Istanbul, Artut says, “a couple people who were reading the text said ‘we can’t understand anything,’” to which he replied, “I don’t either.”
But Artut enjoys the feeling of trying to understand what the work means. “I would prefer to leave it to the audience to analyze the work instead of giving them an instruction manual,” he says. While his work functions as commentary, then, it also encourages dreaming and interpretation. He would rather “have nonsense statements than provide heavy meanings.”
Fundamentally, though, the work is an opportunity to reflect on the all-too-frequent absurdities of complicated art writing, in which adjectives and clauses resolve into nothing but a maelstrom of confusion. While humorous in Artut’s work, this experience has more serious ramifications in its ability to alienate people from the art world.
Complex and obscure language of this sort can, of course, leave viewers feeling unable to comprehend an intellectualized reading of a work. The texts Variable produces, while admittedly not even adhering to grammatically correct English, bring this feeling to the fore:
“We are defining this variable is a transcendens. But have we not presuppose something that only the answer must lie in the sense that it in each case are. Beings are, so to speak, be envisaged in a preliminary way?” one statement reads. “But its systematics is not needed cannot be represented by lower ones. Sciences and disciplines are ways of movement must be sought in the first place can an age lack the discipline of world history is lacking, that is to achieve clarity regarding its own character...” another rambles.
At the same time, Artut’s experiments with machine learning engage with interesting questions about artificial intelligence and creativity: As the artwork titles and statements autogenerate, and Heidegger’s philosophies re-sample themselves and apply them in analysis of the new work—new forms of art emerge.
If a machine can one day “think like Heidegger,” I and many others have wondered, could it eventually create like an artist? And would we call that art? Either way, as far as Artut is concerned, Variable should speak for itself: “I’m not saying this is a masterpiece,” he says, “but if an artwork is a masterpiece, there needs no explanation. That’s why it becomes a masterpiece.”
