GAN to the Mississippi: Generative Adversarial Networks for Posthuman Relations to Place

Keywords: Generative Adversarial Networks; Visual Methods; Mississippi River; Posthumanism
Abstract Type: Virtual Paper Abstract

Abstract

Many have discussed how images construct place, from maps to landscape paintings, images have been a part of defining relationships between people and their environments, often, in the case of the Western canon, as part of colonial and nationalistic projects. As new and emerging Machine Learning imaging techniques are becoming part of representing (or misrepresenting) place, I ask how ML image generators called Generative Adversarial Networks (GANs) build relations to place by making critically informed images of the Mississippi River. The choice of the Mississippi River is connected to longer-term research into how technologies build relations between people and nature, undergirded by ecological posthuman thinking. Ecological posthumanism argues, in the face of climate change, that humans must shift from hierarchical to networked and entangled ways of seeing themselves in relation to non-human others like technologies and ecologies. The Mississippi River is a site of long-term human/non-human entanglement that resists romantic idealization of nature due to its large-scale environmental engineering, historic floods and storms, ties to commerce, and intersectional issues of race, class, and colonialism. Using the lens of ecological posthumanism as a starting point, I probe the limits and possibilities of GANs to create posthuman images of the Mississippi River and ask what this paradigm shift means in how emerging visual media forms build relations to place.