Katja Novitskova There is not only a heat wave at the moment, but also a drought in Europe. Every few weeks there is a new virus or ecological disaster, and it is becoming just a normal situation.
Joanna Zylinska On the one hand we have a drought and on the other hand we are drowning in a flood of data—and it is all quite overwhelming. There is also a flood of disaster news headlines, covering everything from the climate to AI and other forms of apocalypse.
KN This is definitely a good background for our conversation. You wrote a chapter about my work in your book called AI Art: Machine Visions and Warped Dreams. There are a lot of interesting overlaps between my practice and your research. My first question is: how did you develop your idea of nonhuman photography?
JZ Like you, I feel we have been in conversation for a very long time. We have been looking at each other’s work, seeing and sensing the world in similar ways. It is great that we now have an opportunity to exchange ideas and see points of convergence and divergence between us. One of the many reasons I was attracted to your work in the first place was because of your very creative way of working with images and words. I really like how text becomes a form of image in your practice. My book Nonhuman Photography, which came out in 2017, was aimed as a reflection on what is currently going on with images. The majority of images today, as Trevor Paglen points out, are not taken with a human viewer in mind. We are also witnessing a displacement of the gaze from humans to machines. So with this term “nonhuman photography” I meant three things: images that were not of the human, such as depopulated landscapes; images that were not by the human, including devices such as CCTV, drone cameras, telescopes, or medical imaging cameras, which take photographs without direct human intervention; and, last but not least, I was thinking about images that were not made for the human, such as QR codes but also fossils as a form of “proto-photography.” In Nonhuman Photography I tried to show that photography has been nonhuman for a very long time. The first picture in the history of photography, the view from the window from Nicéphore Niépce’s house in Burgundy, took eight hours to produce. It presents a distinctly nonhuman view because there are shadows on either side of the image. In a similar vein, William Henry Fox Talbot described his country mansion, Lacock Abbey, as the first house that took its own picture. We therefore have this nonhuman dimension already at the very beginning of photography.
KN You also have this idea of an image as an expanded entity. It is not just a picture; it is a process of trace-making in a mechanical way. Even Benjamin H. Bratton mentions that photosynthesis is a form of vision because it is a reaction to light. It is a fixation of light in the medium.
JZ Absolutely. This also links with Lynn Margulis’ work on life, organisms and symbiosis, and thinking about how all living organisms perceive. Perception is a key driver of life that functions not just in complex organisms, such as human and nonhuman ones. Perception is also a form of image-taking, of capturing something or, to use Bergson’s terminology, of carving out space from the optical flow. I am trying to expand the notions of image and image-making by going back to early organisms and thinking of imaging as more than just a human practice, and more than a set of technical and mechanical activities. Image-making can actually perhaps be found at the origin of life.
KN In your draft of a new book, The Perception Machine, you mention that “perception occurs in the world as much as it does in the eye and the brain.” For me that means that when light hits proteins in a retina, the electrical signal is already an image encoded that goes into the brain and then expands into a picture.
JZ That is why all these current developments around machine vision are, on the one hand, fascinating and, on the other, disappointing. They are mimicking human vision while using a very simplified, almost two-dimensional idea of human vision, believing that you can reduce vision to pattern recognition and to just seeing edges. Neuroscience, biology and cognitive psychology are all showing us that we do not fully understand vision and perception yet. We know from the writings by psychologist James Gibson, who predicted it before advanced scientific experiment could confirm it, that vision is environmental. Vision happens between the eye, the brain and the world. We can of course trace how molecules travel and see their movement between the optic nerve, the eye and the brain. Nonetheless, we cannot trace the process of the occurrence of images or the emergence of consciousness or intelligence as part of that movement of molecules; we can’t see the behavioral processes which are needed to reflect on and grasp the very seeing of that image. That sense of what it means to grasp an image is unclear to us as yet. Maybe it is only in a dynamic relation that these things occur—and that fascinates me.
KN This dynamism you can see as different in each individual organism, in each individual species, in each individual environment. But what specifically I find fascinating is your contextualization and criticality towards these ideas of pattern recognition and edge detection, and the classic framing of machine vision. You are looking through that or beyond that. I find it very refreshing, specifically in regard to my work. I am obsessed with the use of words such as patterns and approximation, but also with the idea that vision is an approximation. Do you have an idea as to what vision could be, beyond what it is limited to?
JZ Maybe this multiple approximation is already good enough and would allow us to open up from the current model that has been adopted in AI and machine vision research, which is based on David Marr’s famous book called Vision: A Computational Investigation into the Human Representation and Processing of Visual Information. Marr assumes that vision is computational: that it is in itself a reduction and flattening. This model could of course be useful to produce some pattern and edge detection, and to teach algorithms to detect these lines and edges, or rather what we humans see as lines and edges—because other organisms could see object boundaries very differently. So maybe one gesture could be to open up beyond the anthropocentrism of vision. There is not one vision. There are different forms and modes of vision that function differently in different organisms—in terms of distance, perception, depth and angle. Vision also has a visuality, as theorists such as Hal Foster or Nicholas Mirzoeff have pointed out. It is impossible for us humans to think about vision as a purely biological phenomenon. Biology is also a form of culture. Timothy Ingold says biology and culture are only different in terms of time scales, they are not ontologically different. They just unfold at different scales and temporalities.
KN You use the word planetscape to contextualize this problematic vision towards the whole planet. Could you expand on this notion?
JZ I am a little suspicious of this turn to planetarity that is currently occurring both in cultural theory and in the art world. It allows some artists and theorists to retain a certain hubris with regard to their own position. What I am trying to introduce instead—and this is another point of connection between our respective projects—is humor, irony and a certain gaze that comes from askew. The planetscape becomes for me a space for expressing the multi-layeredness of the world, our entanglement, seeing ourselves as part of the world, rather than standing above it as artists or writers. And then recognizing that maybe the only way to deal with that hugeness of scale and the impossibility of capturing it all is through a mixture of affects such as humor, irony, parody and pastiche, all of which can have a generative effect.
KN What you have just been describing is featured in your projects, like Feminist with a Drone, where you specifically talk about “loser images.” Perhaps I am making “loser art” because I am trying to capture a lot of things but not doing it fully and it is a bit amateurish. The effort is there but then the result just falls flat because it is impossible.
JZ There is also a knowingness in what you are doing. There is a certain heroism in it too, almost like “the heroism of the weak,” which is also a form of strength without the hubris.
KN Could you expand on this idea of “loser images” versus “poor images”?
JZ You referred to my project Feminist with Drone, for which I bought a toy drone to capture some images. I was trying to respond to all these amazing images taken with amateur drones. People, especially guys, would visit beautiful locations and capture these amazing views and post videos on YouTube or stills on Instagram, tagged #amazingdroneviews, with motivational slogans, creating this whole beautiful world reduced to a diagram or postcard. I was thinking “okay, let me just try and see what I can do with this new planetary vision developed through drones.”
KN What happened?
JZ Everything went wrong. I lost the drone, then I found it, then the drone lost a propeller and the ability to fly, and then I lost the will to live, with things getting worse and worse. The images that were produced were completely rubbish. But part of me was enjoying the rubbishness and lostness of it all. I was thinking of this idea of “loser images” as a counterpoint to the increasingly perfect images produced with amazing apparatuses. These days any mobile phone takes a sequence of images and averages it to offer us the best one. Everyone is the best photographer. With my project I was trying to respond to these perfect images: not only are they perfect, but they are also taken from high in the sky, conveying an almost celestial vision of perfection. So I was trying to introduce a little bit of dirt and messiness into the picture of the world by presenting my loser images taken with that broken drone, and putting them through a search engine called Same Engine to come up with a whole lot of images from other people, crowdsourced from Pinterest, Facebook and other websites. I also used this approach as a method for doing research, again through irony, building on Polish philosopher Ewa Majewska’s idea of “the avant-garde of the weak,” and on Hito Steyerl’s concept of “poor images.” Your practice also uses these images that come from the Internet. These are images which very often are not original, they’ve already been copied multiple times and jpegged to the point of non-visibility. Looking for imperfections is perhaps a more ethical way of understanding our relationship with machines—including imaging machines. To introduce this idea of “loser images” is also to acknowledge a form of neorealism. There is an ethical dimension to seeing what we call reality in such a messy way.
KN 480-pixel resolution somehow feels more realistic than 8K.
JZ I have got really bad eyes; I have multiple floaters and I had eye surgery when I was younger. Due to this, I have got avant-garde cinema constantly playing in my eyes. But, even if we are talking about human vision, whose vision are we talking about? Does a person with a perfect 20/20 vision even exist? Is there not always some kind of brokenness in the forms of vision, even if your visual apparatus is seemingly perfect? Everyone sees through “loser images,” they are just pretending they are not.
KN Since last year we have also become deluged with AI-generated images from all these algorithms such as DALL·E and MidJourney, that are generating realistic photographic pictures from text or other images. How do you think it relates to vision globally?
JZ I recently asked DALL·E a question: “what will our future look like?” It presented me with a picture of a boy and girl. They are sitting with their backs towards us, and they are looking at a terrain which looks partly bucolic and partly scary. The posing of this question was in a context of a present time, when we are increasingly seeing ourselves being seen, tracked and touched by an ever-growing volume of cameras and scanners. The machinic image apparatus also has a predictive function. It forecasts us into the future by means of images, while playing a certain version of this future before our eyes. I am describing this genre as an “after-photographic hybrid.” It often looks like an Instagram story intercut with a horror film, creepily reimagined by AI. Algorithmically-generated images tend to look like post-apocalyptic film stills—and I am intrigued by that.
KN Do you think that the results of DALL·E depend on what you ask, as if it was a glass ball or a fortune teller?
JZ Of course. It is like the results of your Google search, isn’t it? Or like your adverts on your Instagram, when you ask yourself “how does it know?” And of course it does know… This leads me to think that my ideas are not fully mine. We are talking about nonhuman photography, but we can also think about nonhuman knowledge—and nonhuman intelligence.
KN This is how I also think about my work, like there is almost nothing in it that I draw myself or make physically. It is all like recaptured patterns or forms from somewhere else to the point that, maybe that is problematic. You are editing or filtering everything into an essay or an artwork.
JZ We do that anyway, with words and images. If you are a living breathing individual, you are already part of this data and media flow. You can regulate and can do something with those data points, but it is a fantasy to think we are outside that grid. I am very taken with Vilém Flusser’s idea that a creator today is somebody who in-forms (or “gives form to”) the dataflow. It is a less romantic, more realistic and more honest way of understanding what we call creativity. Your work confirms that: you give up on this idea of the artist as a genius that stands above the world. I think there is genius in your work, precisely in that entanglement: you rearrange, you re-channel those flows.
KN I am just a filter. I think of myself like a sponge organism.
JZ But not all organisms can re-channel things in that way or have a desire to re-channel them!
KN Of course, in the end I still put a name on it. I am claiming some sort of authorship. This question of power takes me to the topic of proteins. There has been a myriad of powerful algorithms released by Google affiliates. One of them has been the algorithm called AlphaFold2, which is a subsidiary of Alphabet Inc DeepMind, a research laboratory which has worked for many years on developing an algorithm to predict acid sequences of proteins. It is an investigation into how a sequence of aminoamides or proteins translates into a three-dimensional shape. Until last year only about twenty thousand protein shapes had been successfully folded and known to science. But, all of a sudden, AlphaFold comes out and they can do millions. Now any scientist can look up a specific protein and see what its model looks like. This is one of these extensions of vision and topics. Approximation of form is a form of vision. This radical vision is a much more complex case of seeing biological structures and all the power that comes with that. There are very few institutions on earth that would be able to make this kind of jump, but, as Google has all these data centers and all this expertise, they were the ones who developed it. On one hand, I am just personally very intrigued by what it opens up. On the other hand, if you ironically type “corona spike protein,” it turns out to be a forbidden thing to ask. I got a notification saying that if I try to do it again, I would be kicked out.
JZ Oh! All right, Google told you off.
KN I did not even know. Of course, I knew there are certain things you cannot ask. But then I had no idea that the spike protein of coronavirus would be a forbidden topic. This is a way to bridge the actual entanglement and environment versus machine intelligence and somehow all the power agents that are in between. I am looking forward to this whole topic being more widely researched, not only by people within tech but also by academia and other artists, because there is something very important in there. I have done a couple of projects, such as Microbial Oasis. What I am trying to do now with this idea of mutation and vision somehow goes into both computational and biological directions. I am trying to connect them.
JZ It really ties in with what I tried to do with my new book The Perception Machine: Our Photographic Future Between the Eye and AI, a follow-up to Nonhuman Photography which is coming out from the MIT Press in 2023. One of the chapters is titled “Can You Photograph the Future?”. I begin it by looking at different forms of research around AI, when they tried to predict the movement of a nematode worm. They could only do that a few seconds into the future but then it all fell apart because the degrees of complexity were too great. There is an infinite number of possibilities, but just a short window of possibility where you can photograph things into being. In Indo-European languages, there is a clear link between images and imagination. You image things into being by imagining them or you imagine by imaging them. What you are trying to show here is like imaging life, opening it up to imagination.
KN You can predict novel structures and by predicting them you are basically creating them.
JZ Absolutely. That is why photography has this agential force, because photographs become living objects. They are not just representations. I have never worked with the model of photography as representational. I think this belief was a form of deceit. Photography has always had an agential force. It has always been able to do things and I think now, with the mobilization of other technologies, it can do them in a much more radical way.
KN And what do you think is the ethical edge in it? You have written about it in Minimal Ethics for the Anthropocene.
JZ My ethical propositions were not principles but rather pointers for ethical thinking. A good approach is not to do ethics too quickly but rather think how we can keep an ethical inquiry open, instead of just trying to come up with a set of rules. Humans have been trying to invent ethical catalogs and decalogues since time immemorial, yet we have not been very successful at implementing them because, obviously, politics and different forms of violence tend to intervene. So for me ethics is not a matter of writing down some life principles but of realizing the socio-political dimension of life. And these are the questions we should be asking: Who has the right to produce life? Whose life is it? Who will be included in and excluded from it? Ethics is a relationship between yourself and the world, it’s a one-to-one relationship. Politics comes into it when there are more than two people involved, and there is a need to make decisions about value, finance, economics, belonging/not belonging. In a way, the most ethical way of dealing with ethical problems is to bring in politics.
KN Yeah. You mentioned that you position ethics above ontology. We are saying that this machine vision, machine creativity, photography in a sense, is a level of ontology. There has to be a level of choice. That choice is the ethics basically.
JZ For me ethics is even stronger than choice. I am developing this idea from philosopher Emmanuel Lévinas: ethics is a form of responsibility that is placed upon every human. Responsibility is contentless. Lévinas understood that responsibility comes from the fact that there are other beings in the world whose existence precedes mine, and who make a demand on me. There is no way of not responding to them. Ethics hits you as soon as you get into the world, and you have to position yourself in it. Figuring out how and what you are going to do with this demand is a lifelong task. The beingness of other beings—human and nonhuman ones (and this is where I depart from Lévinas’s humanism)—already comes with a responsibility understood as something that demands a response from me, from my mind and body.
Loser Images. Soft Approximations
in conversation with Joanna Zylinska
Are We Eternal Beings?
Fall Winter 22-23
Katja Novitskova (b. 1984, Tallinn, Estonia; lives and works in Amsterdam). Her work tackles the complexity and eventual failures of depicting the world through technologically driven narratives. By uniting art and science to the level of nature, Novitskova brings awareness to the mediation and representation tools used to depict these realms. Through her use of AI technologies, she creates works which vacillate between physical and digital domains. Recent solo exhibitions have taken place at: Kraupa-Tuskany Zeidler, Berlin (2022); Kunstfort bij Vijfhuizen, Netherlands (2021); Whitechapel Gallery, London (2018); and Kumu Art Museum, Tallinn (2018). Recent group exhibitions have taken place at Marta Herford Museum, Herford (2022); MUDAM Luxembourg (2021); Belgrade Biennale (2021), and Migros Museum für Gegenwartskunst, Zürich (2020).
Joanna Zylinska is an artist, writer, curator, and Professor of Media Philosophy + Critical Digital Practice at King’s College London. She is the author of a number of books, including AI Art: Machine Visions and Warped Dreams (Open Humanities Press, 2020), The End of Man: A Feminist Counterapocalypse (University of Minnesota Press, 2018) and Nonhuman Photography (MIT Press, 2017). Her art practice involves experimenting with different kinds of image-based media. In 2013 she was Artistic Director of Transitio_MX05 Biomediations, the biggest Latin American new media festival, which took place in Mexico City. She is currently researching perception and cognition as boundary zones between human and machine intelligence, while trying to answer the question: “Does photography have a future?”