Meet the New Kid

‘Machines Enrol in Art Class!’ The title of the American Scientist article (1) I’m currently reading is droll take on the process of artificial intelligence (AI) learning. I imagine the first art class I ever attended had a robot classmate. “I want everyone to be very welcoming to our new student! Class this is DALL-E.”

DALL-E’s name is a clever blend of surrealist painter Salvador Dali and robot character WALL-E. It is the most popular AI art platform in the world, as well as the face of a quickly expanding industry. The purpose of DALL-E is evident in its namesakes: simply, it is a robot that creates art.

Artificial intelligence is described as "the science and engineering of making intelligent machines (2).” More specifically “machines that think like humans.” AI art is an application of this wider machine learning. In short, it is art created by a “thinking” computer. In mid-2022 the world of AI art became a monster of industry, with AI art platform Midjourney reporting over 12 million users since its launch in July 2022 (4). However, as with any quick advancement in technology, there are issues that come with the new power we’ve been given; there is a question we must ask ourselves: how far is too far?

Back in art class, other students and I share uneasy glances as our strange companion clicks and whirrs behind his desk. I smile at him and reach out a hand. He breaks my finger and steals my Snoopy drawing right off my desk. Ouch. This is the first impression many artists had when AI art was first introduced. For years we had been told that truck drivers, factory workers and other industry roles would be fully automated in the future. This was an issue in itself, but an inevitability as the wheels of advancement turned over. Few expected that the creative industries would so quickly slip under the control of automated technology. With AI on the rise, many artists fear for their livelihoods, job prospects and their intellectual and creative property.

To both create and vet the art for our personal viewing experience, AI uses tools called Generative Adversarial Networks (5). Imagine two little robots hunched over side-by-side desks. One’s job is to create images and fool the other into believing its art is original. If the second robot is fooled half the time, the first is performing as it should, creating art the public will believe is genuine. Fool me once. The definition of “genuine” art is vague at best and still under heavy debate. I believe that “genuine” is a feeling you get when you look at the art in front of you. It’s up to you to decide, can AI art ever really be genuine?

Another issue arises from AI art: usually when you go to an art class, the teacher knows they’re teaching. My robot classmate begins to shuffle through its filing cabinet of pre-existing works made by humans. It’s been fed these images, paintings, photographs and learnt to reassemble the input as AI-created art (1,6,7). I can’t help but bristle in contempt as he examines my sketch of Snoopy and adds it to the ever-growing collection of “borrowed” art.

As public use of AI continues to rise, we will inevitably have more ethically grey tundra to cross. In 2018 the Portrait of Edmund Belamy, sold for 432,500 USD at Christie’s Auction house. The controversy surrounding the piece stems from the fact that it was painted by an AI that had been fed 15,000 portraits from 20th century artists (7). Edmond De Belamy is not a real person, nor is the person who painted his portrait. Both are simply amalgamations of the people who came before. Which begs the questions, who deserves to be rewarded for this artistic feat? Is this painting even original? Once upon a time it was something genuine; it is up to us to decide how many times we can take the derivative of a piece of art before it loses its emotion. Or is it simply always the case that every artist takes inspiration from those who came before?

When each creation is the derivative of a piece, it’s critical to examine what it’s deriving from. It is both fascinating and alarming that social biases have even made their way into the creative and malleable minds of machines. Dark-skinned people and hijab wearers have taken to social media to report their selfies, which they had run through an AI software, had returned distorted, warped or whitewashed (8). AI learns from the dataset that its human trainers provide, and it's important that we consider who and what are represented in that training.

It is a strange sort of embarrassment to know that these machines and their endless learning have picked up on our shortcomings. In a world that already rejects difference and shame people for their deviation from unachievable standards, we have taught machines to do the same. Like a child passively taking up its parents' opinions. In this way, some consequences of our technological development aren’t so much AI art issues as they are societal issues. The only fix is to work to include and recognise all people in creative spaces so that machines can learn to think like all humans.

Despite criticism and philosophising, we cannot stop the march of progress. AI will continue to advance, to become better at thinking and emulating us. Perhaps our responsibility will just be to give them something good to copy.

As for traditional artists left behind by the AI advance? I believe we’ll be okay. We differ from machines, and there will always be something more rewarding than just an output. The joy is in the creation, the connection, and the humanity of art.

Half of my art class is robots now, churning out hundreds of creations per second. I hold up a half-finished picture to my desk mate to see their smile. That moment of connection is why it’s worth staying until the end of the class. REFERENCES

Elgammal A. AI Is Blurring the Definition of Artist [Internet]. American Scientist. 2019. Available from: https://www.americanscientist.org/article/ai-is-blurring-the-definition-of-artist

McCarthy J. What is AI? [Internet]. Stanford.edu. 2012. Available from: http://jmc.stanford.edu/articles/whatisai.html

Midjourney Discord Interface [Internet]. docs.midjourney.com. Available from: https://docs.midjourney.com/docs/midjourney-discord

Hughes RT, Zhu L, Bednarz T. Generative Adversarial Networks–Enabled Human–Artificial Intelligence Collaborative Applications for Creative and Design Industries: A Systematic Review of Current Approaches and Trends. Frontiers in Artificial Intelligence. 2021 Apr 28;4.

Goodyear S. Why those AI-generated portraits all over social media have artists on edge [Internet]. CBC. 2022. Available from: https://www.cbc.ca/radio/asithappens/artificial-intelligence-ai-art-ethics-greg-rutkowski-1.6679466

‌Christie's. Is artificial intelligence set to become art’s next medium [Internet]. Christies.com. Christies; 2018. Available from: https://www.christies.com/features/A-collaboration-between-two-artists-one-human-one-a-machine-9332-1.aspx

GANs. Edmond De Belamy, From La Famille de Belamy [Internet]. Caselles-Dupré H, Fautrel P, Vernier G, editors. original gilded wood frame. 2018. Available from: https://www.christies.com/lot/lot-edmond-de-belamy-from-la-famille-de-6166184/?from=salesummery&intobjectid=6166184&sid=18abf70b-239c-41f7-bf78-99c5a4370bc7

AI selfies — and their critics — are taking the internet by storm. Washington Post [Internet]. Available from: https://www.washingtonpost.com/technology/2022/12/08/lensa-ai-portraits/