This presentation proposes a critical examination of generative AI as a medium for subversive, investigative, and open-source practice, arguing that machine learning systems are not neutral infrastructures but contested terrains in which artists can operate as tricksters, infiltrators, and counter-archivists. Drawing on Donna Haraway's concept of situated knowledge, I contend that the partial, positionless view inscribed into AI systems constitutes both a site of critique and an opportunity for artistic intervention: a productive instability to be exploited rather than resolved.
I will present three interconnected works that enact this methodological stance. Beautified-AI (The Photographers' Gallery, 2023) deployed AI-generated portrait fabrications to expose the asymmetric logic of identity construction in datafied environments, implicating GDPR frameworks and surveillance capitalism within an augmented reality intervention. Deepdreamcatcher (Matadero Madrid, 2024) pursued a speculative inquiry into machine consciousness through live-coded soundscapes and generative imagery, destabilising the boundary between tool and agent. My research paper, Restoration and the Use of Variational Autoencoders (VAE) in Algorithmic Rituals (A+E Lab, 2025), theorises generative reconstruction as a form of ritual repair, asking what is lost, what is invented, and who decides, when a model regenerates the past.
Collectively, these works operate within the logic Lucy Lippard attributes to activist art: simultaneously inside and outside institutional structures, working with imperfection and failure as generative conditions. I will argue that subversive AI practice, low-budget, self-directed, and ethically entangled offers a model of research and production that resists the consolidation of machine learning knowledge within corporate or purely technocratic frames, and ask how artists might continue to sidestep, hijack, and occupy these systems.
Leda Sadotti is a creative technologist and researcher working at the intersection of generative AI,computational media, and subversive digital practice. Based in London, she holds a BSc in Creative Computing from the University of the Arts London and currently works as Creative Technology Technician at UAL's Creative Computing Institute.
Her practice engages critically with machine learning systems as sites of contestation, probing fabricated identity, algorithmic ritual, and AI consciousness.
She has exhibited and presented internationally, including at The Photographers' Gallery, Matadero Madrid, The Wrong Biennale, and MACA Amsterdam, and has authored research on Variational Autoencoders in algorithmic practice.