Technology analysts and science fiction writers have long prophesied a future in which increasingly capable machines will play a fundamental role in human society. Yet for many people, the most advanced examples of artificial intelligence and robotics are computer chess games and assembly line automation, both of which have existed for decades. However, the age of thinking machines has just begun, and profound transformations are currently underway. According to technology author Luke Dormehl, artificial intelligence can now “drive cars, trade stocks and shares, learn to carry out complex skills simply by watching YouTube videos, translate across dozens of different languages, recognize human faces with more accuracy than we can, and create original hypotheses to help discover new drugs for curing disease.”
These are remarkable capabilities—but do they constitute thinking? Renowned scientist and mathematician Alan Turing anticipated that computer sentience would become an accepted phenomenon by the end of the 20th century, presaging that “the use of words and general educated opinion will have altered so much that one will be able to speak of machines thinking without expecting to be contradicted.” Although Turing’s timing may have been off, his prediction is increasingly conceivable—particularly in creative endeavors that presumably require more forethought than rote widget-making.
The inventive capacity of today’s machines is the focus of “Artists & Robots,” an exhibit at the Astana Contemporary Art Center in Kazakhstan, on view from June 10 to September 10, 2017. Curated by Miguel Chevalier and Jerome Neutres of the Réunion des Musées Nationaux-Grand Palais, “Artists & Robots” dares to ask questions such as “Could a machine do what an artist does?” “Could a robot replace a painter or a sculptor?” and “To what degree can we talk about artificial creativity?”
The eye-opening collection of 17 works includes in-process projects like “BeBot” by conceptual artist Leonel Moura. BeBot consists of a swarm of small wheeled robots that roam a large horizontal platform, each marking the canvas surface with a different color pen. To the observer, the robots appear to be drawing at random, collectively creating a field of haphazard lines reminiscent of a geometric Jackson Pollock. However, a much more sophisticated underlying process is at work.
Although the robots begin making random marks on a blank surface, they quickly shift to a responsive mode upon detecting lines of their own color. Over time, a stochastic approach is gradually replaced with an intentional one: “This means that when the canvas starts to be filled with color the random behavior stops,” Moura says. According to the artist, BeBot exemplifies stigmergy—a zoological concept that Oxford defines as “The production of behavior that is a direct consequence of the effects produced in the local environment by previous behavior.” Stigmergy is a critical process of cognition, as outlined by scientists Leslie Marsh and Christian Onof: “To know is to cognize, to cognize is to be culturally bounded, rationality-bounded and environmentally located agent.”
The display of something resembling consciousness is startlingly evident in Patrick Tresset’s “Human Study #2, La Vanité.” Devised as a “theatrical installation” according to the artist, the work consists of an immediately recognizable collection of objects: a still life with a human skull, seashell, and other artifacts, surrounded by several small wooden desks. Yet this is no traditional drawing class. In place of a human artist, each desk is attended by a mobile camera and stylus. These devices—a cybernetic eye and hand—are physically attached to armatures mounted to the desk’s surface, and digitally connected via computer to work together. In a highly methodical and coordinated process, the camera swivels to inspect various aspects of the assembly of artifacts, and the pen arm then generates marks corresponding to the light and shadow appearing within the camera’s visual field.
The effect is nothing short of astonishing. The pen is not merely plotting prescribed information, but recording what the camera eye “sees” in real-time; meanwhile, the machine is somehow making judgments about pattern density, part-to-whole relationships, and the moment the drawing is complete. Because each robot has a different vantage point, its drawing is distinct from the others—and because the process includes inherent variability, each picture is unique. Each machine’s drawing style “is not a pastiche but rather an autonomous interpretation influenced by the robot’s qualities and faults,” states the artist.
“Artists & Robots” culminates in the most portentous work of all: Memo Akten’s “Learning to See.” Viewers enter a space in which three walls form an immersive projection chamber. The first wall consists of a matrix of blurry images and a running line plot. This projection represents the first phase of “training” for an artificial deep neural network—the kind of system employed by sophisticated surveillance organizations. The computer is simply confronting new imagery and trying to make sense of it. The second wall features a matrix of image samples from the Google Art Project, which includes scans from art repositories around the globe. Here we see various scenes and abstractions flickering and shifting as the AI attempts to codify discernible patterns. On the third wall, viewers observe the machine attempting to create its own art based on what it is learning. A grid of imagery vibrates with ghostly depictions of landscapes, figures, and spaces—all drawn by computer, akin to a kind of surrealism-meets-abstraction.
Akten is tapping into the phenomenon of increased computational power and large repositories of information—known as Big Data—while interrogating how machines can make sense of it. “But what does it mean to ‘understand’? What does it mean to ‘learn’ or to ‘see’?” he asks. Akten reveals the process of machine-learning in real-time, from a phase in which computers attempt to decipher the work of humans—to one in which we try to understand their creations. In this final stage, we are the ones who are learning.
So what do these works suggest for the art world—and by extension, the creative disciplines? According to Dormehl, we first need to ask what it means to be creative. Obviously, it is not sufficient simply to make something; it must also be novel—not only within its context but “new to society as a whole.” As designers are well-aware, such novelty typically arises not out of thin air, but by introducing a new interpretation within the repository of known standards and precedents—otherwise called knowledge. Computers are already adept at mining this great database, and increasingly, interpreting it as well.
In the future, designers may call upon machines to analyze project briefs and begin working on the design itself, while they provide occasional guidance as the deep learning algorithms form lines, shapes, and volumes that satisfy a project’s complex circumstances. Or perhaps we may work alongside machines, watching them formulate new schemes based on our own process sketches or models. Either strategy will likely result in an increased number of design options as well as a more detailed assessment of project constraints. However, working with machines in this way will also involve more effort, as we transition from being mere users to operating as stewards and partners of thinking technology. Eventually, such a relationship may enlighten us about the nature of design itself—an activity that is habitual for creative professionals, yet shrouded in mystery.
The spelling of artists Leonel Moura and Memo Akten's names have been updated. ARCHITECT regrets the error.
