AI Through the Viewfinder: Reimagining the Camera as a Tool for AI Image Generation

Abstract

The rapid emergence of artificial intelligence (AI) is causing profound shifts within the art world, reigniting age-old debates on the boundaries of what can be considered art. For example, many AI systems are employed to mimic the styles of existing artists and their works. Although this approach is deemed to be derivative and uninspiring to many people in the art world, it is also forcing us to reconsider longstanding beliefs attached to creativity such as the importance of originality and authorship. Given that AI is here to stay, this thesis explores a critical question around AI and perception, asking “How and what does AI see? Specifically, this thesis investigates the types of biases that are ingrained or embedded into AI systems, and how these biases are reflected in the output, specifically in the context of images. As part of this investigation, this thesis culminates with a prototype - an AI camera that embodies the process of AI ‘seeing the world’. This camera integrates photography with artificial intelligence, serving not only as a tool for technical exploration but also as a metaphor for examining how AI technologies offer diverse and potentially transformative perspectives on reality, much like a traditional camera. By abstracting AI technology into a camera, this project aims to start a conversation about how AI, like a camera, offers us different, sometimes biased views of the world. In doing so, the camera is redefined from a mere tool for capturing images to one that generates them, and in some cases (mis)represents human forms and identities.