How queer artists are resisting generative AI

For many queer artists, making art is a way to find community, express joy, resist oppression and preserve history. But as generative AI becomes embedded in creative platforms, some artists are changing how they share their work. 

A new study led by University of Michigan School of Information research fellow Jordan Taylor explores how queer artists understand, refuse and resist generative AI. Through interviews with 15 artists, the researchers found tensions between participants’ community-centered artistic practices and the ways AI technologies are developed and used. 

"I Just Don't Want My Work Being Fed Into The AI Blender: Queer Artists on Refusing and Resisting Generative AI” was published in the September issue of the Proceedings of the ACM on Human-Computer Interaction. 

“Queer art is often not about how we can optimize our performance and make as much money as possible,” Taylor says. “These are communities that are built around caring for one another.”

The artists interviewed described several ways they resisted generative AI, from choosing not to use it in their own work to encouraging others in their communities to avoid it. Some also stopped sharing their art online because they feared it could be collected and used to train AI models. That decision meant giving up opportunities to connect with other artists.

Not every participant’s relationship with AI was straightforward. Some appreciated the strange, surreal images produced by earlier generative AI models, a quality Taylor connects to an appreciation for strangeness and failure in queer artistic practice. Taylor points to “The Rocky Horror Picture Show” as an example of a film whose strangeness has helped it resonate with queer audiences for decades.

“There’s something so kind of beautiful in its very strangeness,” Taylor says.

Participants were more critical of newer AI models that aim for realism, Taylor says, because they lost some of what made the earlier images interesting.

The findings also encourage researchers to think beyond a familiar response to AI bias: adding more diverse material to training datasets. For artists who object to how generative AI is developed, greater inclusion in those datasets may not address their concerns.

“In order to make sense of that harm, you have to think about the importance of art within queer communities,” Taylor says. “Both as kind of a tool of resistance, as well as a tool of collective memory and joy and collectivity.”

Taylor joined UMSI as a postdoctoral research fellow after earning a Ph.D. in human-computer interaction from Carnegie Mellon University. At UMSI, Taylor works with associate professor Nazanin Andalibi to study the labor implications of voice AI, including how it may affect voice actors.

Read “I Just Don't Want My Work Being Fed Into The AI Blender: Queer Artists on Refusing and Resisting Generative AI” in the September issue of the Proceedings of the ACM on Human-Computer Interaction.