Could thoughtful ‘AI art’ counter the rise of ‘AI slop’? Why creatives today face a paradox
A discussion on Reddit includes a photo of a poster advertising the upcoming triennial at the National Gallery of Victoria. The poster shows an image from Ayoung Kim's multi-screen video artwork, Delivery Dancer's Arc: Inverse (2024).
The image shows a Korean woman wearing a riding suit and gloves as she runs through a crowded, futuristic city. It comes from a video about women who work as delivery drivers for gig-economy companies in a dystopian version of Seoul. In this world, artificial intelligence forces the women to compete against one another, watches their work and distorts their sense of time.
The artwork combines AI, computer-generated imagery (CGI), video game engines and live-action footage.
The comments on Reddit strongly criticize the use of generative AI. They do not distinguish between art made with generative AI and situations where AI replaces the artist completely. They question why the gallery chose to advertise a work that presents AI as an artistic tool.
There is an irony here. The commenters object to a world where AI takes over human work. But Kim's artwork also explores this very problem.
The debate seems to depend on her choice to use generative AI as a tool for making art.
AI everywhere
AI systems shape our everyday lives. They are used in warfare, influence what we see in our social media feeds, analyze our facial expressions and take data we upload so that a small number of technology companies can profit from our cloud capital.
Our data is used to build the datasets behind commercial generative AI platforms. These platforms fill the internet with what people call “AI slop”: large amounts of AI-generated content. And we seem to have no way to opt out.
It is understandable that Reddit commenters are upset to see art made with generative AI. If AI also replaces art and creativity, what parts of social life will remain untouched by its spread?
However, their criticism treats two different things as if they were the same: AI replacing an artist and an artist choosing to use AI as a tool.
So, does an artist's intention matter?
Media theorist Joanna Zylinska is careful to distinguish between two kinds of art made with generative AI.
She compares popular AI art to a mind-numbing game of Candy Crush. This kind of art uses generative AI only for what it can do, celebrating its ability to turn data into eye-catching images.
Critical AI art, by contrast, helps people see the effects of the technology. This includes how AI systems are built, how they affect human work and how they may change our understanding of what it means to be human.
The role of the gallery
Art institutions around the world are paying more attention to how AI affects art and creativity.
Some of these artworks strongly criticize how AI systems take and use resources, including data and human work.
But art institutions also depend on company sponsorship, visitor numbers and media attention to secure funding. As a result, an artist's political criticism can become a marketing tool, an icon or a main promotional image.
Ayoung Kim's AI-generated poster shows this problem when it is taken out of the context of her artwork.
Theorist Victor Burgin questions whether art galleries can effectively criticize AI. He argues that to understand how a work of art criticizes AI, people need to know its background. They cannot understand the full criticism simply by looking at the artwork.
For Burgin, art can resist harmful ideas in other ways: by challenging stereotypes, clichés and ideas that have become deeply rooted in popular images and visual culture.
Understanding AI through art
As people become more concerned about AI replacing artists' work, and AI spreads into more areas of the creative industries, it is important to discuss what artists can learn by experimenting with this technology.
As a creative researcher, I am interested in how my experience of a technology shapes what I can understand about it. As a photographer, I think about how generative AI developed from datasets containing photographs from around the world. This also means that generative AI inherits ways of seeing and remembering the world that come from photography.
Generative AI appeals to people's desire to create images of almost anything, from nude images and Hollywood stars to pictures that make it look as if dead people are alive again.
When text-to-image platforms appeared in 2021–22, people could turn these desires into images by entering just a few words.
That is the promise made in the technology's marketing. It led me to create a work about something that was fundamentally impossible to see: my mother before she became my mother.
I wanted to find out whether the billions of images taken and used without permission could produce even one image of my mother.
The experiment was designed to fail because of the prompt I used. I did not want to solve this contradiction. Instead, I wanted to use the problem to explore the relationship between visual language and the labels attached to images within AI systems.
Since making this work, I have continued to think about the ethical problems involved in using generative AI as an artist. AI slop is one reason I hesitate. The political questions around whether we can refuse to use AI are another. I keep asking myself whether artistic practice can help us understand and think critically about AI systems.
Making art with generative AI can begin by using these systems in new ways and working against what they are designed to do. This means accepting what AI cannot capture and challenging the way it works. For example, artists can use prompts to explore parts of social life that are difficult to measure or combine into data using algorithms, such as feelings and physical sensations.
In this way, experimenting with art may be one of the few opportunities we have to experience the limits of AI as it increasingly spreads into our lives.