Inexistent Archive

Abstract

An inexistent archive is a series of fictional photographs of gay couples and queer or non-binary, working-class people set in early 20th century Latin America. The images were created with generative artificial intelligence, so they do not correspond to real people. It is a retro-futuristic exercise that uses computational algorithms -technology usually immersed in futuristic narratives- to reimagine our local queer/kuir past and reclaim an archive that could not even exist. As Cuban-American critic José Esteban Muñoz explained, one of the consequences of heteronormative culture is that the queer experience of the past could leave almost no record or archive. This denial of the archive is even more dramatic for people from the global south and the working classes. Initially, the text discusses some background of works that experimented with generative AI entitled Artificial Imagination. Additionally, the article points to some “techno-poetic” considerations and consequences of working with AI models. For example, those deriving from the use of the reverse “diffusion” process, which could be described as cheating the model, forcing it to return to a hypothetical prior image that never existed. Or the need to counteract the class and race biases present in the databases on which the models were trained, an obstacle from which the concept of “minority prompt” emerged. Finally, the article also explains the decision to maintain the errors in body representation that are inherent to these techniques. These errors are integrated into the project as an additional layer of bodily dissidence and as an ethical-political limit that evidences their contrived origin, to prevent these images from covering up the past of violence that prevented their very existence.