JAMES COUPE
art projects
Warriors: Generative Adversarial Network
Categories: Art

Exhibition: Exhibited at International Center of Photography , New York City,
January 23 – May 17, 2020
Materials: 5-channel video monitors, computers, iPads, cameras
Description: This work uses a scene from Walter Hill’s 1979 cult movie The Warriors, which involves an an encounter between an all-female gang, the Lizzies, and three members of the Warriors gang who are lured and then attacked, but eventually escape. Female characters in The Warriors are depicted as strong, determined, and often aggressive, in contrast to some of the male characters.

Generative Adversarial Network uses AI classifiers to profile visitors’ faces based on a variety of gender-related markers, and then uses a customized version of the notorious Deepfakes algorithm to switch visitors’ faces into the film. Faces swap at a variety of speeds and resolutions, and do not attempt to be seamless in all cases.

Deepfakes uses a generative adversarial network (GAN), which means two neural networks battle each other in order to try to find a solution that each is satisfied with. Both the original scene in the film, and this deepfakes interpretation enact themes of gender fluidity and performance by switching and swapping between male and female faces and reversing roles of aggressor and target. Viewers are assigned roles within the scene based upon how they are classified and measured by the profiling system.

Credits: With thanks to the Center for Digital Art and Experimental Media (DXARTS), University of Washington. Software development by Jacob Fennell, Forrest Fabian Jesse and Yuying Hung.
Reviews/News: 15 Perspective-Altering Shows to See in New York City in 2020
IMAGES: GAN GAN GAN GAN
VIDEO:  WarriorsVideo

Categories: Art -