On the Parallel Struggles of Photography and GAN-generated Imagery
On the Parallel Struggles of Photography and GAN-generated Imagery

Summary/Abstract: This article analyzes certain aspects of the reception and acceptance of the emerging technology of GAN generated visuals (DALL-E, Midjourney, and Stable Diffusion) by the general public as of mid-2021 to early 2023. It will deal specifically with the negative and positive expectations of it and put them into the context of social acceptance and expectations surrounding the formation of the photography medium in the early 19th century. Both of the inventions mark a milestone in the timeline of a data-driven society. Both tools utilize a reference to reality or its visual coding. GAN image generators, as well as photography, were expected, as a product of science, to deliver almost mythically perfect, objective outputs. While photography had previously been used in pseudoscientific projects to support racist claims, the public appears surprised by the bias of the GAN datasets, and their creators are currently looking for ways to eliminate it through process moderation. The goal of using this comparison is to explore a common trait of both epochs: the affinity for devices, specifically the idea of “a black box”. It will aid us in describing the relationship of the data society to itself based on machine-based creation of visuals with varying degrees of autonomy and generativity. Finally, we will examine the practical implications of the discussed aspects in the near future.

• Details
• Contents