How Artificial Intelligence is Changing the Arts and Democracy

In celebration of Human Rights Day, 11 December 2023 witnessed at the mdw an event that revolved around the arts, artificial intelligence, and democracy. Its keynote, “The machine’s obviously not in the mood* – Wie künstlich (intelligent) ist Kunst und was bedeutet das für die Demokratie?” [… – How artificial(ly intelligent) is art and what does this mean for democracy?], was delivered by Eugenia Stamboliev, a University of Vienna-based doctor of musicology and philosopher of technology.

When the BBC documented the first electronic music at the Computing Machine Laboratory in Manchester back in 1951, the recordings they produced seemed downright playful; at that time, nobody could’ve dreamed that computers would go on to cause much change.1 And nearly 80 years later, we’re still wondering: Can computers create art? Particularly the new AI platforms raise this question anew. Fed with endless reams of data and artworks, their systems can create algorithmic motifs in seconds. This ability casts doubt upon the notion of authorship and blurs the boundaries between work and product. It also demonstrates the power of large corporations such as OpenAI, DALL-E, and Midjourney, which reserve the rights to their results even as they dominate visual and democratic norms.

Eugenia Stamboliev © Stephan Polzer

Computer art was something experimental. So why is AI now painting Rembrandts? The artist Vera Molnár demonstrated how computers can represent an interesting medium all the way back in the 1970s, though this took quite some time to be acknowledged. So is it just that we need time to recognise AI’s significance as a producer of art? On the other hand, AI art currently seems to have other problems. Amidst a repetitive loop of aesthetic norms, it’s become difficult to distinguish between AI art and advertising images. Midjourney and DALL-E are most prominent among the image-generating platforms contributing to this chaos. Such platforms process instructions or prompts with AI in order to produce new motifs. A problematic aspect of this is that they draw upon data from countless millions of sources, frequently in violation of applicable copyright laws. In doing so, they produce a recognisable AI aesthetic that sells well since we recognise it quite easily as art—and here, Rembrandt clearly trumps Malevich.

The AI portrait Edmond de Belamy by the artists’ collective Obvious is a good example of how effective the patina of age can be. In 2018, it fetched over USD 400,000 as the first AI picture to go up for auction at Christie’s. Today, most AI-generated visual artworks drip with artistic pathos and a level of kitsch that frequently undermines both art and AI as such. These trends represent neither art’s democratisation nor a creative explosion of AI, but much rather a battle for a market share2. There are also serious debates ongoing that reflect upon AI’s methods—such as concerning the promptography of photographer Boris Elgadsen or video works by the artist Claudia Larcher. Aside from all this, the debate over the aesthetics of image-generating AI isn’t necessarily what’s truly interesting: many art and media scholars in fact ascribe greater creativity and value to such algorithms’ material and performative aspects.3

All the while, AI aesthetics are popular, prejudiced, and frequently undemocratic. A search for “family picnics” on Midjourney reveals motifs that seem straight out of a 1950s mail-order catalogue from OTTO-Versand and have little to do with a modern patchwork family. Those who seek to distort or alter such outcomes indeed can, but one soon finds out that increasing the degree of aesthetic distortion (accomplished by turning up the “weirdness” level) causes things to suddenly appear not just more impressionistic—like the portrait of Edmund Belamy—but also more diverse, as pictorial scholar Roland Meyer notes. While that may sound good at first blush, it isn’t. All it means is that image-generating AI defines people and families who are of colour, diverse, and/or non-heteronormative as “weird”. What may initially seem more aesthetically playful and inclusive is, in fact, democratically intolerable. One can also see this when using English-language prompts. Searches for “scientist” or “doctor” turn up almost exclusively old white men. Such wrongs have already been described at length4—but instead of changing, this type of old-fashioned patina and conservatism has also now begun appearing in political campaigns.5

Escaping Old Data Sets and Norms

AI must not be allowed to throw us back intellectually and artistically. We’re already so far beyond it, aren’t we? On the other hand, just like Midjourney’s AI-generated conductor (“composer”) is still an old white man, so is the conductor of the New Year’s Concert—which would suggest that AI is frequently just a good mirror. Altering stereotypes first requires changes in real life, and institutions such as the mdw (where 50 percent of conducting students are now female) make enormous contributions toward disrupting these patterns. Such disruption requires commensurate will, time, effort, and resistance to being set back by AI systems. Here, art—with and without AI—can accomplish a lot by way of new spaces and narratives and through new protagonists.6 After all, technologies like AI always enable us to do both: to prop up the status quo and/or to shine a light on its blind spots.

Let’s Hack AI

Technology, democracy, and art are closely intertwined, but how do we deal with AI’s prejudices and dependencies? We can attempt to carve out more freedom for ourselves, but in what way? According to Vilém Flusser, it’s not just the case that we programme media: media also programme us. We adapt to their possibilities, in the process forgetting that we can also do differently—that we can repurpose a medium, disrupt it, play with it, or even hack it.7 It is only in this way, holds Flusser, that we can assert our freedom—particularly when it’s artistic freedom that’s at issue. So if we really want to understand AI, we must find ways to address those aspects that we find undesirable in a holistic manner: technologically, artistically, and democratically. This will take no small social, political, and artistic commitment, but it’s a price that we must be willing to pay.

  1. (accessed on 20 Dec. 2024).
  2. (accessed on 05 Dec. 2023).
  3. Luciana Parisi, 2013, Contagious Architecture.
  4. Caroline Ciado Perez, 2019, Invisible Women.
  5. (accessed on 06 Jan. 2024).
  6. Joanna Zylinska, 2020, AI Art: Machine Visions and Warped Dreams.
  7. Vilém Flusser, 2014, Gestures.
Comments are closed.