Historically, 'trophy hunting'—the practice of hunting wild animals and collecting them as trophies—is intricately linked with colonialism. Due to the characteristics of its training data, AI deeply internalizes human cultural contexts and notions. As a result, the images AI generates of 'trophy hunting' vividly retain the colonial aesthetics of such photos—depictions of white men in dominant poses against a backdrop of twilight savannas. Notably, rifles are learned with such strong weights that they are difficult to dissociate from these images.

The work specifically focuses on this phenomenon and undertakes an intentional experiment to remove rifles from the AI-generated images, observing the tensions that emerge during the process. The artists continuously input various opposing concepts, attitudes, entities, and objects to counteract the generation of rifles, trying to delete or alter them. However, AI persists in generating rifles or leaves their traces according to learned statistical patterns, sometimes even creating images that defy its own set policies. Nonetheless, AI consistently believes that 'no rifle has been generated.'

The 'Muscle Data,' imbued with deep human cultural histories, becomes tense amidst the artists' conflicting requests, leaving behind peculiar glitched artifacts or traces mingled with rifles. The artists name these ironic images and the traces of rifles 'Botched Trophy.'

​

*Muscle Data: A term coined by the artists to denote data that has been learned with strong weights in AI, similar to muscle memory. These data are difficult to remove with deliberate prompting and tend to generate automatically, representing data deeply embedded with cultural and historical contexts.