In the Generative AI Art Boom, Who Wins and Who Loses?

In 1763, a Presbyterian minister, Thomas Bayes, laid the groundwork for machine learning when he developed a framework for reasoning about the probability of future events. A couple of centuries and billions of dollars later, we have text-to-image AI systems that are changing the way we approach art forever. While generative AI art produced using diffusion models has been around for a while (the AI-generated portrait of Edmund Belamy, created by artist collective Obvious, sold at Christie’s in 2018), there has been a visible increase in public interest in AI art with the launch of mainstream programs like Prisma Lab’s Lensa AI and OpenAI’s DALL-E. The appeal of quickly creating unique images and avatars from infinite numbers of text prompts, with a technology that improves with every use, has drawn the attention of millions of people.

The success of these AI models—from the web-based DALL-E to Midjourney,  available exclusively through Discord—has meant an increase in prestige, interest, and funding for their parent companies. Midjourney was used earlier this year to create the front cover for an issue of The Economist and OpenAI has received over a billion dollars in funding from Khosla Ventures and Microsoft. 

This is not without a ton of controversy,  the most prominent of which was an image created using Midjourney, which won first place in a digital art competition in August. The image, “Théâtre D’opéra Spatial” beat out 18 other artworks and earned its owner, Jason Allen, a $300 prize. While Allen defended his use of Midjourney—on grounds that the use of such AI image generators wasn’t explicitly disallowed by the competition’s rulebook—other artists on the internet saw it as the beginning of the end. “We’re watching the death of artistry unfold right before our eyes—if creative jobs aren’t safe from machines, then even high-skilled jobs are in danger of becoming obsolete. What will we have then?” a Twitter user wrote.

For all the newfound hype benefitting influential creators, the generative AI platforms they use, and, of course, the financial backers positioned to be the biggest beneficiaries of this particular market boom, a group of people has been a little more skeptical about this technology: artists. 

AI art is a relatively new concept, at least to the general public, and with it comes a set of new questions and conversations to be had: if generative AI purports to be the “new norm” in creative fields, who are the winners and losers as this trend plays out? 

AI image applications like Midjourney, Stable Diffusion, and DALL-E are trained on millions of images including images by artists without permission and credit. This is a big concern for artists including Greg Rutkowski, a digital artist whose name is frequently incorporated into prompts to generate AI art. In an interview with MIT Technology Review, he discussed his apprehension with AI art and, more broadly, his take on the future of art. 

These conversations on copyright, ownership, and AI art have resulted in a broader question of attribution: whether art produced by generative AI models is more human or more machine. On the one hand, these images are created entirely by generative programs, but these programs were trained on a corpus of work from real human artists, to whose contributions it’s currently difficult (if not downright impossible) to attribute credit . While it is difficult to determine who owns what now, it might be harder in the future when these lines are blurred even more

Many commercial artists fear that AI art generators are going to replace them. It doesn’t help that these generators are quickly growing in popularity because of how fast and cheap they are. Realistically, AI won’t completely eclipse human artists anytime soon; at the very least, people will always play a role in curation, selecting the AI-generated images which best match their creative objectives. 

Despite the prevailing discourse, there are some artists and designers who are embracing AI art and are using it to revolutionize their creative workflows and their perspectives on creativity writ large. Artists like Mario Klingermann and Helena Sarin are using artificial intelligence to create thought-provoking art that might not be possible without the help of AI. Another artist, Memo Akten, produced a film constructed entirely through the latent space of a deep neural network in 2018 and regularly gives speeches on the intersection of art and AI. 

There is a rising interest in prompts for generative AI as a form of art since it sometimes takes very specific prompts to get very specific results. As prompt engineering and other new artistic forms related to AI art are growing in popularity, so are the number of artists who cater to different parts of the AI art creation process, as they come to terms with, adapt, and participate in this emerging phenomenon. 

For other people, the appeal of these applications are often much more personal. Programs like Lensa AI and Midjourney are allowing people to memorialize deceased family members, document important milestones in their personal lives, and recreate historic moments in their families. Other people are simply using it to create unique avatars for themselves. Lensa Ai, which creates stylized portraits and avatars from photographs of users, has gone viral on Tiktok and Twitter because of all the ways it is being used to create images that are unique to a person, and is currently at the top of Apple’s App Store charts in the U.S..

With AI art, the possibilities are endless. Users can literally create new worlds from virtually any era or style desired. Regardless of where one  stands, this is groundbreaking work that should be celebrated. Yet, that doesn’t take away from the legitimate fears and concerns of the people whose original work was used to help educate the AI program in the first place. That AI art is the future is unquestionable, but the direction of that future depends on how much human artists are harmed along the way