TroublingGAN
generative artwork, installation, 2021-22
TroublingGAN is a critical generative artwork and artistic research project that experiments with the StyleGAN vector-image generative neural network. It experiments with alternative use of GANs as a tool for artistic research.
TroublingGAN is a custom-trained StyleGAN︎︎︎ model that generates a visually ambiguous images depicting "troubling times". It is able to derive an essence from a dataset representing a "troubling times" - and project the affective quality of photojournalism from the dataset onto the generated outputs. The motivation for creating this seemingly imperfect generative model was to arrive at a new type of understanding of the vague notion of "troubling times" using a neural network perspective.
The project uncovered a number of issues related to visual representations of disturbing events and disasters and the affective value of contemporary photojournalism. Such journalistic photography is being often re-used in different contexts, which causes troubling unethical use of sensitive visual material as illustrative stock photos. This project argues against recontextualization of journalistic photography and suggests speculative use of visually ambiguous generated imagery instead. Replacing the once-photojournalistic-new-stock-photo with generated semi-abstract visuals coming from a neural network that has learned from similar thematically identical photographs, this ethical problem ceases to exist and space is created for a different way of perceiving the image.
Such images do not burden the viewer with unnecessary meanings and the frequent information noise does not arise. They are devoid of context, but still carry the necessary atmosphere. But despite the absence of specific objects and scenes, these visually ambiguous images are still strangely reminiscent of photography. This is very confusing to the human eye. The mind is constantly trying to assign some meaning to these obscure compositions, however abstract. But the assigned meaning or interpretation becomes dynamic and constantly changing, and therefore it is ultimately the atmosphere and emotional charge that affects the viewer. Thus, neural networks, through their interpretation of the dataset, offer a different understanding of the affective quality of journalistic photographs.