Exploring generative AI and its ecosystem.

Time has passed since DALL-e broke the news when it showed to the world that images could be generated from a text fragment such as “An astronaut riding a horse in photorealistic style”. Now the technology of AI image generation has proliferated in a large offer of interfaces, softwares, scripts and is distributed in forums, social networks, in the news and in art galleries where all kinds of imagery are flourishing from unicorns and space ships to fringe porn and deep fakes. It raises many questions regarding authenticity of information, the automation of creativity, the democratization of art, the propagation of gender and racist biases. So where does one start to try and understand this phenomenon? And what are its implications for image practitioners?

In this workshop, we will try to explore the ecosystem of Stable Diffusion, a set of open source technologies and communities that support different platforms of AI image generation. Following various maps of the generative AI infrastructure, we will look into the material and economic dependencies of this system as well as its social dynamics, its ecological footprint and aesthetic perspectives. We will also experiment with its different components and its image sources (datasets) to sense the range of possibilities it offers. We will prompt, code, submit images, play with models and delve into the – sometimes obscure but never dull – world of LoRAs, refiners, upscalers and the communities of freaks and enthusiasts that power them.

Nicolas Malevé is an artist, programmer and data activist. He is currently a postdoc researcher at the
School of Communication and Culture at Aarhus University, supported by a grant from the Novo
Nordisk Foundation (NNF21OC0068539). His work explores various modes of intervention in the
politics and aesthetics of computer vision.
Here you can find Nicolas Malevé’s personal page on the website of Aarhus University.

Organized by Prof. Anne Quirynen (Potsdam University of Applied Sciences).