![](https://www.klemms-berlin.com/content/uploads/2023/10/viktoria-binschtok_21-angry-young-woman-2023-framed-1152x1446.jpg?x83594)
Viktoria Binschtok43% happy
Wolfgang Ullrich /
Viktoria Binschtoks “Typewriter Photographs”
At first, we feel like we’ve been transported back into the early 20th century: Photos of an old mechanical typewriter, zoomed in to show the type hammers, with a white sheet of paper, covered in type, clamped to the carriage. The lower-case ‘a’ is a bit out of line, and the type has often punched through the ribbon, making the letters look messy. The pages contain only individual words or numbers, sometimes arranged in rows, sometimes printed half on top of each other, sometimes with large spaces between them. The overall appearance is reminiscent of ‘concrete poetry,’ harking back to a genre from the 20th century, when poets and artists experimented with setting words in constellations and creating meaning in ways that went beyond syntactic relationships.
The title of Viktoria Binschtok’s exhibition–“43% happy”– brings us back to the present. Quantifying emotions by percentage is a typical feature of AI-assisted facial recognition programs, which are trained to detect ‘action units,’ that is, individual states or processes in a person’s facial expressions, and to relate them to each other. The percentages arise from the fact that the combination of ‘action units’ leads to different forms of emotional expression in each case. A person who is only smiling, for example, is identified as less happy than a person who also has laugh lines around their eyes.
In addition to the emotional state, recognition programs survey a face according to other parameters, for example drawing conclusions about a person’s age, gender, or ethnicity. In this case, the program identifies the individual who is 43% happy as a 28-year-old woman with brown hair. Viktoria Binschtok has given the photograph on which this information appears the title “Mona L.”– making it clear that this brief, matter-of-fact description applies to of one of the most famous–and most photographed–works of art in the world. Binschtok’s piece shows what remains of the Mona Lisa when a facial recognition program evaluates it according to a few, sometimes vague, sometimes perhaps even misleading categories, and one then types out the result on a mechanical typewriter.
The other photographs in the series also present doubly transformed versions of images. The character of these images varies: some are press photos, others are private snapshots or advertising images, but we now see them all only through the categorizations of the AI program. Closer inspection, however, reveals that she apparently did not always use the same image recognition software. What is verbalized and how changes depending on the interests of those who use such an app. To find a person wanted by the police in a crowd (as in “suspect found!!!”), one must train the AI program using multiple images of the same individual.
By contrast, for AI to identify trees, dogs, or cars in addition to people (as in “Sunset Boulevard”), one must first provide the app with a very large, heterogeneous body of images. Millions of images have to be described individually–tagged by hand–before the self-learning systems can detect subjects in other images in a reasonably reliable and detailed manner.
In general, image recognition software involves much more human piecework behind the scenes than we would expect from high-tech. Cheap labor prepares this material, often under barely tolerable conditions; not much has changed from the mindless work processes characteristic of the early days of industrialization. Binschtok’s “Typewriter Photographs” make this palpable. By tracing image recognition processes back to mechanical procedures and directing the viewer’s gaze to the metal hammers of the typewriter–we can almost hear a frenetic, loud clacking when looking at her photographs–she demystifies the supposedly clean and noiseless application of digital programs. The rather messy appearance of the letters and the uneven constellation of the words reinforces this effect.
Viktoria Binschtok presents her series at precisely the right moment. As AI programs become increasingly popular and more and more people use them, we must ultimately learn to better understand their prerequisites and consequences. This applies not only to questionable approaches to training these programs, but also the enormous amount of energy they consume, and issues of data protection. By transferring the results of AI applications into a mechanical, analog environment, however, by using this form of alienation, Viktoria Binschtok also prompts viewers to think about the negative effects of digital technology and the extent to which it permeates all areas of our lives. As in her previous series “Clusters” and “Networked Images,” she successfully develops a concept that is as formally independent as it is incisive, in order to raise awareness of the fact that images in the digital world become data and can thus be used in a variety of ways.
Not only that: Just as Binschtok combines analog and digital worlds, blending the 20th and 21st centuries in her series “Typewriter Photographs,” she also creates a new space that resists attempts to locate it within time. We feel removed from our own present moment, and when we ‘read’ the typed images of information captured by AI, it sets off a mental cinematography that leaves everything open. Reflecting on and analyzing media thus simultaneously becomes–first and foremost–poetry.