08.07.2024
Jonathan Luke Austin
On 27 and 28 June 2024, HUD’s Jonathan Luke Austin and Maevia Griffiths attended the two-day symposium Synthetic Vision/Images of Power at the Framer/Framed platform for contemporary art, visual culture, and critical theory & practice in Amsterdam. The event hosted talks and discussions that explore the transformations induced by Artificial Intelligence (AI) within the interplay of power, knowledge and images. Specifically, it focused on the unprecedented possibilities of AI in generating ‘synthetic vision’ (the ability to “see” algorithmically) and ‘synthetic image generation’ (the ability to create new images through prompts). Full details of the event can be found here.
At the event, Jonathan Luke Austin and Maevia Griffiths presented work from a longer-term collaboration on the possibility of deploying ‘sousveillance’ technologies for violence prevention in different contexts. While expressed in this context in terms of its conceptual and theoretical coordinates, this work draws on earlier work by Austin on sousveillance technologies applied to preventing torture in detention and on wider humanitarian crises. An abstract of the talk is below.
Synthetic Vision for Subversion: Thinking Beyond the Algorithmic
What’s the future of synthetic vision? Can it be imagined beyond the algorithmic? As a subversive mode of worlding reality? To explore that question, this presentation meditates on three cases of synthetic vision that offer starkly contrasting visions of its normative potentials. First, we introduce a research project being developed with the support of Chinese security agencies. The project involves developing machine vision techniques to detect abuse in detention and policing settings by authorities. It works to reduce human beings – police or detention authorities – to pixelized skeletal forms interpreted by an algorithm, erasing any specificity to their personhood or the ecologies they inhabit. Purportedly, to improve compliance with legal norms. In a second move, we compare this project to Wassily Kandinsky’s series of line drawings Dance Curves in which he reduced the form of a dancer he observed to a series of skeletal lines remarkably similar to that of the algorithmic vision of Chinese security agencies. In the Bauhaus tradition, he sought to minimize – reduce – the form of dance and the body to abstraction, erasing personhood. In a third move, we explore one filmic representation of the process of becoming a torturer, entitled Grievable/Ungrievable. Grievable/Ungrievable is an abstract representation that meshes the movements of a dancer’s body with her descent into the materiality of an underground facility. Without voice, or narrative, a process of becoming violent emerges. An experimental engagement with film and dance as a mode of revealing the embodied essence of violence. We discuss this project as a form of ‘artistic’ synthetic vision in which the moving images produced seek to represent a form of image-truth and image-evidence but without the restraints of algorithmic or technical mediation. Overall, these three acts of minimizing human form and subjectivity are put into deliberately provocative tension to help think synthetic vision in the full scope of its normative ambiguity to meditate on alternative political trajectories. Specifically, we draw on these examples to problematize the dominant intuition that what Latour termed ‘irreductionism’ (avoiding such minimization of human form and subjectivity) is always normatively preferable to the reductionist nature of the kind of technical mediation epitomized by synthetic vision technology. Instead, we speculate on the possibility of designing a subversive synthetic vision.
