Jonathan Luke Austin
Professor
The University of Copenhagen
jla@ifs.ku.dk
Maevia Griffiths
HUD PhD Researcher
The University of Copenhagen
mlg@ifs.ku.dk
Across the world, government, commercial, and other actors are creating forms of ‘synthetic’ vision. Typically, synthetic vision is used to monitor people walking in the street – to, for instance, detect criminal activity. Or, more dramatically, by drones to identify potential threats to be targeted. Or, in more banal terms, by self-driving vehicles to navigate the streets. But can synthetic vision be imagined in more emancipatory terms? As a humanitarian tool or process? In this short essay, HUD’s Jonathan Luke Austin and Maevia Griffiths reflect on that possibility conceptually. They do so through three examples that seek to expand the repertoire of what synthetic vision is usually understood as constituting. Placing those examples in uneasy tension, they then suggest a turn towards a more experimental ethos in humanitarian politics, and beyond. A version of this essay was presented at the 2024 Synthetic Vision workshop in Amsterdam by Austin and Griffiths.
This short reflection forms part of two core HUD sub-projects. The first, led by Austin, focuses around developing ‘sousveillance’ (under-surveillance) tools and technologies in humanitarian settings. Such technologies, as the essay suggests, might reverse the gaze of surveillance technologies by observing powerful actors in ways that might reduce their tendency to deploy violence. The second, led by Griffiths, focuses on how visual practices can ‘move’ people towards more subtle and politically sensitive forms of intervention in humanitarian and violent contexts. As expanded on below, that project explores – in particular – the virtues of experimental forms of film-making and visual practice to transform humanitarian communication.
1/
Violent-Dancing
What do you see in this blurry, red-white-black image?
What does it mean?
What does it do to you?
These questions will elicit only vague responses because you have not seen this image in its original form, as a moving image: a pixelated video-clip. In this still-image form, it is unknown, unrecognizable.
To make sense, it needs a story.
Somewhere in Northern Syria, sometime around 2013, a strange kind of violence occurred. In a seemingly abandoned building, a group of soldiers from the Syrian Army had begun beating – eventually, to death – several men lined up against a wall.
This is a still image from one of several videos recorded by the soldiers carrying out this act.
It is then an image of violence. But again it’s an image that, in this form, is unrecognizable, something that becomes an object of uneasiness only when you are told what is. But you still see very little. Except pixels, blur, colour. This image, and its story, is a representation of something that seems ‘unreal’ to our lives. It is a piece of evidence, it reveals that ‘something’ happened. But it offers little truth. It is a synthesis of something ‘over’ reality (sur-réel).1For more on this discussion see Della Ratta, Donatella. 2018. Shooting a Revolution: Visual Media and Warfare in Syria. Pluto Press.
Indeed, the story behind it is also sur- real, over- real.
In the process of this beating, leading to death, of these bodies, the mood in that room in Syria changed. A few more smiles become visible on the faces of the soldier, some laughter echoes. The men in the room, with a grotesque melee of kicks, begin linking arms and doing so to the rhythm of the traditional Arab dance the dabkeh. What the Palestinian novelist Emile Habibi termed the poetry of moans and sighs, the poetry of violence: rhythmic yet brutal, emerges quite literally in those grainy images, creating something like a blood painting.2Habibi, Emile. 2010. The Secret Life of Saeed the Pessoptimist. London: Arabia Books.
2/
Violent – Choreographies
So, dance and violence. The disciplining of human subjectivities has always involved controlling bodies in ways that adapt, adjust, and contort their movements.3Foucault, Michel. 1979. Discipline and Punish: The Birth of the Prison. London: Penguin. And that process is not just about the confinement of the body in a literal sense. It is also about controlling its quotidian rhythms, as well as injecting distinct rhythms that enable violence to emerge almost automatically, thoughtlessly.4See inter alia Grossman, Dave. 1996. On Killing: The Psychological Cost of Learning to Kill in War and Society. New York: Back Bay Books. and Wacquant, Loïc. 2004. Body & Soul: Notebooks of an Apprentice Boxer. Oxford: Oxford University Press.
But how do you tell that story?
Consider the now moving images above.
Again, these images are unknown to you at first glance. But they ‘tug’ at something. Make you uneasy. Make you concerned. But you are uncertain. They come from an experimental cinematic attempt to tell the story of how violence emerges, very often, ‘automatically’ or ‘thoughtlessly’ in conflict settings through dance.5See http://www.viftlab.com for more details. The story of how extreme violence has become so embedded in human society that it appears to emerge without reflection.6See Austin, Jonathan Luke. 2016. “Torture and the Material-Semiotic Networks of Violence Across Borders.” International Political Sociology 10 (1): 3–21. and Jonathan Luke Austin. 2023. “The Plasma of Violence: Towards a Preventive Medicine for Political Evil.” Review of International Studies 49 (1): 105–24. But they tell that story without narrative – no dialogue, no explanation, no clarity. And they blur the distinction between the perpetrator and the victim in an uncomfortably embodied contortion of violence, a palpable uneasiness. Unreality.
In all this, those images attempt to abstract the process of ‘becoming violent’ through those automatic processes of disciplining the rhythms of bodies. In this they try to follow what McNeill has described as the “system of command” that choreography imposes on the bodies of dancers – a form of obedience requiring the dancer’s body to be a tabula rasa, receiving the will and corporeal attributes of someone else.7McNeill, William H. 1995. Keeping Together in Time: Dance and Drill in Human History. Cambridge, MA: Harvard University Press. The will of someone else through the repetitive embodied rhythms, the chassé, plié, échappé of this ballet of violence, mimicking the orchestrated movements for our bodies to become part of a militarized status quo.
Unlike that Syrian blood painting, these moving images thus do not serve as evidence to any particular set of concrete facts – people hurting people in a specific place and time – but instead are an attempt to cinematically synthesize a form of ‘truth’ about our world: a reduction of how the choreography of bodily rhythms generates violence globally, affecting all involved.
3/
Playing-Violent
The unruliness of the body, then, is disciplined through the regulated choreography of its movements. And, beyond the individual body, this logic is also central to contemporary military deployments of synthetic vision. Increasingly, security practitioners seek to capture the rhythms of sociality; to pattern them, correlate them, coral them into that which can be captured/controlled. One sees, this, naturally in the perversity of Israel’s use of Artificial Intelligence to destroy Gaza, something echoed in the United States’ use of similar technologies to freeze-frame the vitality of life in Pakistan or elsewhere through its deployment of armed drones.8See Bethan McKernan and Harry Davies. 2024. “‘The machine did it coldly’: Israel used AI to identify 37,000 Hamas targets.” The Guardian, 3 April 2024; Human Rights Watch. 2024. Questions and Answers: Israeli Military’s Use of Digital Tools in Gaza. Human Rights Watch September 10 2024; and Chamayou, Grégoire. 2013. Théorie Du Drone. Paris: La Fabrique éditions..
Within this mix – however – there are other, less dramatic and so less noticed uses of algorithmic synthetic vision.
Consider the images above.
These frames depict typical ‘action-recognition’ algorithms at work. These are tools that reduce the body to a skeletal form, attempting – again – to minimize its complexity and capture what it is ‘really doing.’9For a good introductory discussion of these technologies see Anton Maltsev. 2023. “Action Recognition in the Wild,” Medium, March 20 2023. Sometimes, these capture pedestrians walking in the street, to manage traffic. Other times they are hoped to be able to capture individuals acting violently in street protests, street fights, or other settings.
But there are other, even more ambiguous, uncertain, uses. In several research laboratories around the world, groups of researchers sit in front of cameras and play-out violent scenes. They play-out beating people, play-out putting people in painful stress positions, play-out the minutiae of violence in detention.
They play at being ‘bad’ police officers or interrogators.
The cameras watching them synthesize these enacted scenes into skeletal forms like those above which are algorithmically ‘seen’ (or not) as indicating abuse in detention settings. Reducing state-perpetrated violence to bodily form. But this project, to create a form of synthetic vision that can recognize state violence – automatically, computationally – is being carried out not by human rights organizations, defenders of liberty, or any other such actors, but – rather – at the behest of unnamed Chinese security agencies in the name of increasing the “judicial civility” of its institutions.10See Wang, Hailun, Bin Dong, Qirui Zhu, Zhiqiang Chen, and Yi Chen. 2023. “Spatio-Temporal Attention Fusion SlowFast for Interrogation Violation Recognition.” IEEE Access 11:103801–13. https://doi.org/10.1109/ACCESS.2023.3316724.
4 /
Reducing Violence
So, dance used to enable state violence, dance used to represent violence, and an algorithmic choreography of bodily movement to interrupt state violence.
Across each of these illustrations, unrecognizability, uneasiness, and unreality are ever-present. Each of these impressions is a negative – an un-something. These three stories are all about the reduction of reality.
The underlying ethos in this ‘minimization’ of reality is linked to Hito Steyerl’s principle of the uncertainty of the documentary, the idea that “the closer to reality we get, the less intelligible it becomes.”11Hito Steyerl. 2011. “Documentary Uncertainty,” Re-visiones no. 1. To get too close to the reality of committing extreme violence is to risk admitting its un-intelligibility – its meaninglessness – and so to be unable to carry out that act. So, you dance while you kill. Reduce the act to something else, something joyful. To get too close to the minutiae of the social, political, material, ecological, etc. drivers of such extreme violence in attempting to explain it, intellectually, is to become paralysed by that complexity. To fall into a depressive aporia. So, one reduces it to one particular aspect, bodily form. One reduced to the base-level of its emergence: depicted cinematically, hoping to tug as an ‘affective sparkplug’ towards a different form of understanding. To get too close to the fall gamut of reasons why that euphemistic ‘judicial civility’ does not exist – institutional inertia, political power, economic inequality – is to risk a critique of the state that pays your bills, to risk a self-implication in that violence. So, one reduces the issue, again, to bodily form, mediated through algorithmic pixel points.
We place these reductions of reality in uneasy, unreal, synthesis to open up the multiplicity of their political meanings. Each of these scenes reduces violence in its complexity, its polyphony, its horror. But their politics is different. What we want to ponder now is whether we can reduce violence, quite literally, by meditating on the political disjunctures between these scenes.
In doing so, we now want to twist the idea of synthetic vision. To do so, we can take – simply – its etymology. The synthetic refers to acts that combine, bring together, multiple things in ways that compose them as a singular thing. Only later did the notion begin to refer to uniting things in a regular patterned and predictable structure through, nowadays, computational tools: the synthetic as a grid, a black-box, something post-human. But thought beyond the algorithmic, the etymological roots of the ‘synthetic’ as a concept referring simply to ‘bringing together’ different things into any type of collective composition opens up the possibility of imagining a distinct – less technical – aesthetics for synthetic vision.
How so?
The use of synthetic vision, particularly in military and policing settings, sometimes purports to be about technologically mediating-materializing social rules. The dream that, say, international law can be enforced algorithmically. That the machines will impose the rule of law, and its goodness. That in Richard Brautigan’s classic counter-culture poem, we will be All Watched Over by Machines of Loving Grace.
Irish Murdoch critiqued such a view long ago, writing that the “insistence that morality is essentially rules may be seen as an attempt to secure us against the ambiguity of the world” through an aesthetics of control, capture, and prediction.12Hepburn, R. W., and Iris Murdoch. 1956. “Symposium: Vision and Choice in Morality.” Proceedings of the Aristotelian Society, Supplementary Volumes 30, 50. The idea that we can predict patterns in social reality, seeing it governed by eternal rules, is a defence mechanism.
Against this, Murdoch writes that when we are faced by moral dilemmas:
What is needed is not a renewed attempt to specify the facts, but a fresh vision which may be derived from a “story” or from some sustaining concept which is able to deal with what is obstinately obscure, and represents a “mode of understanding” of an alternative type. Such concepts are, of course, not necessarily recondite or sophisticated; “hope” and “love” are the names of two of them.13Hepburn, R. W., and Iris Murdoch. 1956. “Symposium: Vision and Choice in Morality.” Proceedings of the Aristotelian Society, Supplementary Volumes 30, 51.
A fresh vision or – rather – a fresh synthetic vision, then is required to open up ways of grappling with the “ambiguity of the world” through a distinct aesthetic lens. But how do we get there? For, Gregory Bateson the concept of aesthetics refers simply to the generation of “patterns that connect” parts of the world, in all its multiplicity, that otherwise might seem disparate.14Bateson, Gregory. 1979. Mind and Nature: A Necessary Unity. Dutton. Aesthetics, then, is about story-telling.
So, to get to a different kind of synthetic vision – a story that is not about “specifying the facts” – but instead a sustaining concept that generates different “patterns that connect,” we now want to turn to one final illustration that allows us to re-shuffle, re-mix, re-compose the “patterns that connect” the illustrations we began with. An illustration that takes us away from violence.
5 /
Patterns that Connect
In 1926, Wassily Kandinsky collaborated with the German performer Gret Palucca, to produce a series of line drawings called ‘Dance Curves.’ Kandinsky sought to capture the “simplicity of the whole form” of Palucca’s movements – her body in motion – through very minimalistic drawings: reductions of the complexity of the body.15Funkenstein, Susan Laikin. 2007. “Engendering Abstraction: Wassily Kandinsky, Gret Palucca, and ‘Dance Curves.’” Modernism/Modernity 14 (3): 389–406. He saw this as a way of ‘minimizing’ the subject to access something about its ‘essence.’ A form – then – of synthetic vision situated within the modernist Bauhaus goal of generating a ‘comprehensive work’ of understanding the social-material world. It was a rationalist dream, but also a utopian and emancipatory one. And, so, was eventually declared degenerate by the Nazis.
We have then four images. Depicted below, side by side.
For synthetic vision to have the subversive potential to reduce violence, rather than algorithmically naturalize it, the challenge is to re-shuffle, re-order, re-compose the patterns that connect these images, and images like them. The dilemma that synthetic vision poses in the contemporary period is that the patterns connecting these images are most dense, most entangled, between the first – that blood painting – and the third, that state-funded bid for ‘judicial civility.’ Each enforces the power of the state further. But the other two are nonetheless partially connected to the other two. Dance-as-art is used to enable dance-as-violence. Modernism-as-emancipation influenced modernism-as-pacification.
But how can they be entangled differently?
To tease an answer, we want to conclude now by advocating for what we’ll call an ethos of experimentation.
The cinematic depictions of violence and militarization explored earlier are intensely experimental, characterized by a departure from traditional narrative structures common to experimental cinema in general, which often runs without plot or dialogue, favouring sensory exploration. The genre plays with motion-paintings, visual music as a form of synaesthesia that hopes to make the audience feel like their ‘wires are crossed’, associating feelings that usually don’t come together. Synthesizing different ideas. Generating if not exactly image-truths then image-partial-truths.
Kandinsky, for his part, was also engaged in the experimental practice of synthesizing vision through plain shapes, zig zags, bare lines, and traces of the body. By mobilizing the simplicity of the pencil lines and angles, he paused Palluca’s dance moves – forcing you to hold your breath a few seconds – uncovering the stillness in movement, the pull and push tensions of the body, prioritizing style over the subject matter, highlighting the abstract rather than the figurative and challenging the gaze through minimisation, for a sensory synthesised vision.
One notes – however – that the other two illustrations are also experimental. The Chinese-government funded bid for ‘judicial civility’ is an experimental one, a playing-with the algorithmic to open-up a different kind of ‘skeletal’ synthetic vision to that envisaged by Kandinsky. Equally, that blood painting in Syria was experimental in the simple sense that it was not pre-planned, not orchestrated in advance, but a moment of ‘trying something’ in the moment, connecting embodied patterns of joy and violence.
Distinctions between these illustrations and their experiments are thus ultimately political. And the challenge of moving towards a synthetic vision for subversion – for reducing violence – lies in not seeing their technical content (artistic, computational, embodied, etc.) as representing a qualitative difference. Each of these illustrations are minimizations, reductionisms, simplifications that nonetheless tell specific stories. In gazing at these acts of minimization, from the flesh of social complexity to the bare skeleton of what enables that complexity, we are left with uneasiness about how we see truths. By looking through the body, and by ‘reducing’ the body and its movements to lines and shapes, we perceive ambiguity, patterns/algorithms of violence, as well as its capacity for subversion. As Steyerl states about documentary uncertainty, the blurriness and ambiguity from the proximity to violence, “reflect the precarious nature of contemporary lives as well as the uneasiness of any representation.” Thus, the unease, and uncertainty about the truths of images are inherently closer to reality than the actual claim they are attempting to convey. In this, the challenge pace much art theory, is to think the aesthetic beyond its capacity to “complicate” and instead to experiment with its capacity to simplify and reduce in ways with subversive potential. This requires the difficult ethical task of avoiding the temptation to judge the politics of synthetic vision a priori and, instead, to embrace the more delicate task of experimenting with its different mediums, entangling them differently, with the hope of prefiguring a different kind of political future for its operations.
A montage of Grievable/Ungrievable produced by Maevia Griffiths, with César Brodermann (dance), and Alexane Poggi (dance). A project by the Visibility for Transformation Lab (VIFT). See http://www.viftlab.com .