guest curator series

Vectors by

Amanda Wasiliewski

Photo of Untitled by Laura Owens / Illustration by Aureliano Santa-Olalla Valero and Ana Frías
The big reveal has arrived for this three-decades-long experiment known as the internet: we have finally amassed enough data to form a viable training set for artificial intelligence. All our knowledge and all our junk can now be automatically recycled forever until model collapse. All our myths and metaphors can be mixed, recursively.

The utopian promises from the ’90s are but a distant memory. What were they? I can hardly remember and it doesn’t matter anymore. The internet has long been a dumping ground—the landfill of contemporary culture. “Nothing, Forever”. Facebook may have wished to “give people the power to build community and bring the world closer together”. But, in reality, we have merely connected our trash, brought it all together in its very own vacuum. “AI Seinfeld Broken, Maybe Forever”.

That things are algorithmically generated is nothing new. This machine-made stuff has long lived among whatever remaining meaning there is to be found online. What is new is its velocity and its magnitude, how we set it off in all directions. How we vectorize it.

A vector is a carrier. Of disease and other things. And so, it follows that the internet meme was not the democratization of creativity that some thought but rather the homogenization and standardization of expression.

Even a global pandemic could not reclaim the phrase “going viral”. Instead, we retreated further into vector space. Far away from the messiness of the world, constrained to mathematically-defined shapes.

In 2002, Rem Koolhaas flipped the term “space junk”, the extraterrestrial garbage we leave in orbit around our planet, to coin the term junkspace for its parallel here on Earth i.e. the terrestrial accumulation of cultural garbage that surrounds us.

In 2022, junkspace became latent space. Latent space is the matrix of associations--of vectors--among datapoints in an AI model. We left it up to the machines to sort out our hoarder house of culture.

Aesthetically I was drawn to vector-based programs, the same way I was drawn to Barnett Newman.

Jeff Elrod

Barnett Newman’s famous work Vir Heroicus Sublimis means “man heroic sublime”. The grandness of Newman, the simplicity and purpose of abstraction in the mid-twentieth century, the seriousness can never be regained, drowning as we are in junk. The vir- in virtual originally suggested virtue, potency, and power. Newman’s were perhaps the last virtual artworks. Or it’s a tie for this distinction with Mark Rothko.

There is no longer any possibility of a Rothko Chapel. How can we even fathom that level of earnestness? Earnestness requires a hierarchy. There’s not only no longer any distinction between high and low. There is no high and low. We couldn’t hope to reassert such distinctions if we tried.

“Young people everywhere have been allowed to choose between love and a garbage disposal unit. Everywhere they have chosen the garbage disposal.” The algorithmic trash chute takes our Garbage In, and other algorithms continually return Garbage Out to us. We eat it up. But we did not choose it. Not exactly. There are few of us left who are not addicted to it.

Even abstraction is cultural detritus now. Contemporary artists looking to cleanse the spirit of the junk that surrounds us have no claim on abstraction. Or, rather, abstraction no longer has any claims on purity. Abstraction is just more junk--but this time, vectorized.

Elrod describes his artmaking method as “frictionless drawings”, using computer-based methods and then transferring them to physical space. In physics and mathematics, a vector has magnitude and direction. A force is a vector. The marks in Elrod’s works are identifiable as the neatly subtractive marks that can be produced with computer painting applications. They are then brought onto the surface of the canvas, a once-hallowed ground in abstract painting that no longer can shut out the world or avoid cross-pollination of media. In other words, even the abstract is not abstract. These marks are space junk floating through space. None of the resistance of materials, of atmosphere. Only mass and velocity sustaining endless movement through a vacuum. In cleansing the
surface, this work loops back around to embrace the junk.

Laura Owens’s works also utilize this neat computer-based mark-making. The lines and layers evoke the virtual space--no, vacuum space--of digital drawing applications. Like Elrod, Owens uses these marks for both adding and taking away. Unlike Elrod, she excuses her use of “painting software--like early kids’ paint programs in the ’90s” as a modest exercise in colour. Not a “frictionless drawing” but printmaking. Each layer is a new space, however, and each mark is a vector. So, these works move significantly beyond printmaking and its constraints. Inside the graph of vectors is a vacuum. No texture, no surface, just fill.

“Rabbit is the new beef” begins Junkspace. WOW These 14 Charcuterie Fails Will Make Your Skin Just Fucking Crawl reads the title of a painting by Morgan Blair from 2019. It is one of the shorter examples of the sometimes paragraph-long, humorous riffs on internet textual junk that accompany Blair’s semi-abstract paintings. The series of works she created immediately before these more abstract works was based on screenshots from Seinfeld episodes, a source of inspiration for her humour. Although Seinfeld was a typical ’90s network sitcom endlessly syndicated in reruns, Blair’s works based on them represent a form of viewing only available to those of us in the twenty-first century: computer-based viewing. No longer couch-potato-viewing but bed-rot viewing: people today fall asleep to the soothing sound of reruns auto-playing on a laptop. Nothing, forever.

The choice of Seinfeld is fitting since Blair is an artist who is not interested in earnestness. In an interview from 2018, she explains the long titles of her abstract works with unapologetic cynicism: “In 2013, I made a series of totally abstract paintings. I immediately became aware of the palpable boredom they were causing. My solution was to give them titles like Socially Conservative Man in Conversation with Gay Cousin that suggested the blobs contained specific narratives and references to pop culture that people are interested in”. The narratives may be thin at best but they have magnitude and direction. They are set off into space and likely won’t stop.

The marks in the paintings of these three artists are unavoidably those of Photoshop, Microsoft Paint or whichever of the classic computer art applications you choose from the 1980s to today. The neat swipes are a hallmark of such a raster “painting” program. The gridded transparent “background” void of Photoshop is visible in Blair’s work. All three artists seem to paint with the “erase” tool. If they erased De Kooning, they leave no trace. Each mark is both a mistake and a gesture. Fundamental to any painting program is the ability to undo, to take steps back, to systematize the process itself. Translated into painting, such marks become paradoxical--a meticulously simulated appearance of unseriousness.

The real world has no undo function. Gerhard Richter said: “Photography has almost no reality; it is almost a hundred per cent picture. And painting always has reality--you can touch the paint.” He thus fought against the virtual without being able to avoid reproducing it. As in the Pygmalion myth, painting this virtuality--the artist’s touch--somehow breathes life into it. Richter’s spaces are real spaces masquerading as virtual ones, just as those of Elrod, Owens, and Blair are (David Summers, Real Spaces, 2003).

These works defy the moment--they are both nostalgic and contemporary at the same time. Though they skeuomorphize the material world--imitating analogue ink blots, paint splotches, spray paint--they are something very different. They are fundamentally alien to painting.

Microsoft Paint, despite its name, has very little to do with the tradition of painting in art history. Western art has long debated the virtues of drawing versus painting in art, that is, line versus colour. This debate played out for centuries in Europe: notably, Poussin versus Rubens, Ingres versus Delacroix, the Impressionists versus the Post-Impressionists. Raster images are neither colour nor line. Fundamentally, of course, they are made up of little squares called pixels. Each pixel is a unit--a number, really--of a colour. In combination, they can compose a line but the line seems to deny their essential form: discrete and stagnant. They have no inbuilt direction, no movement.

This is why, when one enlarges a rasterized image, it becomes pixelated. There is no further information inherent in the pixel.

Raster and vector are almost always positioned as opposing paradigms of computer imagery. The vector image-object is composed of mathematically-defined forms. It is made of points, curves, and relationships between them. They are graphs masquerading as form. Computer images started in the vector space, as it requires far less memory to store a vector image than a bitmap. The bitmap, however, allowed a kind of flexibility to make individuated marks. Or at least it gave the illusion of this.

For computer images, the line versus colour dichotomy is irrelevant since raster cannot be understood as pure colour. Contained as it is in a matrix, it has an in-built geometry. Trained in AI systems, even pixelated images can be understood as vectors: machines can be trained to understand images based on pathways forged or determined by gradients, allowing them to automatically “read” images through their pixel transformations.

The use of “brush” and “line” tools in raster image programs complicates the bitmap/vector divide further. The “line” tool in a raster program may simply activate all the pixels in a row to create the semblance of a line but it aspires to be a vector and any illusion of depth is almost always thwarted by the radical flatness of the medium. Despite these output formats, brush and line and various other mark-making tools are vectors in the sense that they are carriers, they have no texture or surface, no geometric unpredictability. They are truly frictionless.

Clement Greenberg could have only dreamed of such flatness in painting. The flatness that this kind of vector formalism allows. Ironically for poor Greenberg, it was Donald Judd’s machine-produced boxes that led the way rather than even the flattest of color field painters. Vector formalism is not only about lines and curves but also about flatness, two-dimensionality. Helen Frankenthaler’s washes of colour or Morris Louis’s curtains of paint could do nothing about their materiality, but Photoshop layers are perfectly flat. Roy Lichtenstein painted the flatness of the printed image, but there was yet a further flatness to come in the form of the computer image. The catch is that it is always an impure image--stacked with junk. It is Leo Steinberg’s “flatbed picture plane” steamrolled flat.

Formalism continues to be a dirty word in contemporary art. The label Zombie Formalism that caught on for abstract painting in the 2010s implies that formalism will forever be Greenbergian. Flatness, forever. The “casualism” and “slacker abstraction” labels for this same work, on the other hand, fail to see that there is no seriousness or earnestness left to opposite in our junkspace-cum-AI dataset. There is no real seriousness to pit casualness or slackers against. And so, there’s just the peeling away of endless, frictionless layers.

In 1949, two French artists wandered the streets of Paris looking for inspiration when they happened upon a hoarding covered with posters for concerts and other advertisements that had been successive pasted on top of each other and torn by both passers-by and the elements. Fascinated by this modern readymade form, they decided to rip off pieces of the hoardings and add them to their painting. This technique is known as décollage, an antonym to collage meaning to take away (unglue) rather than add (glue). Raymond Hains and Jacques Villeglé called their first décollage work made that year Ach Alma Manetro after the snippets of printed text that were still visible from the torn-away posters. In other words, the text and the title are purely nonsense--they’re junk. It sounds almost like Latin but it’s not. It’s the low culture Vir Heroicus Sublimis.

Following this, the work of Elrod, Owens, and Blair is a kind of digital décollage. It might be called a parody of Greenbergian abstraction--if parody could still seriously exist. Décollage is the evil twin of Abstract Expressionism. Digital décollage processes all the junk with none of the friction. Rather than a material torn away from hoardings, rather than physical layers, it is an endless tear through immaterial layers. Each perfectly flat, with a depth of zero. Nothing, forever, indeed.

Amanda Wasielewski is a researcher and artist based in Stockholm, Sweden. She is Associate Senior Lecturer of Digital Humanities and Associate Professor of Art History at Uppsala University. Her recent research focuses on the use of artificial intelligence techniques to study and create art, with a particular focus on the theoretical implications of AI-generated images and formalism. Wasielewski is the author of three monographs including Computational Formalism: Art History and Machine Learning (MIT Press, 2023). Her work can be found at