Is AI art an oxymoron? Can artists and AI coexist? And, what does the proliferation of AI programs in the cultural sphere mean for the future of digital art?  

More and more, these questions are flooding the social media feeds of anyone with even a peripheral involvement in the art world. Digital artists, who make their living online, are seeing their portfolios swiped and repackaged by indiscriminate technology, and they are letting everyone know about this new and unsettling trend that is turning many art makers into advocates. 

Most contemporary AI programs are trained on LAION-5B, a non-profit, open-source dataset. It contains almost six billion text and image pairs and was originally meant to serve as a representation of the connections between language and imagery on the Internet. Now, the dataset is used by for-profit companies to teach their AI how to generate content by familiarizing it with this expansive image and text database. Any user can feed the software a phrase like “Make me an Alexis Franklin,” prompting it to produce an image with an uncanny resemblance to the artist’s decades’ worth of work. Franklin, of course, would have no say in this process.

Recent analysis shows that most images included in the LAION dataset come from Pinterest, DeviantArt, and Getty Images. It is impossible to opt out of LAION—even if you’re the owner of a given image; it’s a picture posted without your permission; or, as is most relevant here, it’s your copyrighted artwork. LAION has even been known to include medical records in its image roundups. 

Apps like Lensa AI, the photo-editing tool that produces fantastical portraits from users’ selfies, are quickly becoming the hottest fads online. Anyone with a few dollars on hand can be rendered as a princess, or alien-like creature. But the technology used is built from Stable Diffusion, an open-source machine learning model trained on the LAION-5B dataset. In other words, the open-source database has been repurposed by Lensa’s developers, charging you $3.99 for 50 AI portraits generated using, in part, art its makers did not consent to sharing with a third-party commercial app. 

Artists do not receive compensation or royalties from the generation or social media distribution of images based on their style and works. Disciples of such AI programs argue that they are challenging the art world’s gatekeeping practices and “democratizing art”—but many artists whose work is on the line feel that the debate is instead about data privacy, intellectual property, and the matter of who is allowed to profit from whose work. 

It’s not unimaginable that in the future, developers will craft AI with a distinct set of perspectives, opinions, and preferences, much like humans. But right now, AI is in its infancy. We can quantify the pixels that these models process, something that is far more difficult to do when it comes to human artists. That doesn’t mean we won’t be able to quantify the processes of our own cognition in the future. Likewise, the flaws that exist in AI right now may disappear as the programs keep training, developing, and expanding. 

However, the road from fledgling technology to ethical and professionally reliable tool is a rocky one. Artists have, rightly, recognized that the current landscape is built upon their labor, however unwillingly they were pulled into the fray. What does that dynamic mean going forward, when more developed technology is available? Perhaps, as the sensationalism that is driving an AI boom fades, artists will have an opportunity to redirect the market in a way that advances both the form and their interests.

This is not necessarily a matter of artist versus AI. Instead, artists have the opportunity to embrace the possibilities that AI offers, treating it like another tool—or even a collaborator. Those most familiar with the risks and pitfalls of the technology are perhaps best positioned to advance it responsibly. Ultimately, artists are the ones on the frontlines, persistently demonstrating the distinction between human and machine. Like the advent of PhotoShop in the 1990s, AI could accelerate the potential for artists to remain on the cutting edge of new creative technologies, while pushing them to develop still more personal, creative perspectives. That’s the one thing AI doesn’t have, yet—Personality.