Last year, in the Colorado State Fair, a man submitted an AI-generated digital image into an art contest and won. This understandably caused significant uproar in the artistic community which claimed that the lack of originality and innovation in artificial digital art should have disqualified the work from the competition. It’s a perfectly reasonable objection, considering most of us don’t even have a concrete definition for what constitutes art—or ‘good’ art, even. But, sitting here, looking at my TikTok feed saturated with ‘dark fantasy AI’ images, I’m starting to see the appeal.

The first of these images I encountered are the two below created by TikTok user shuj1nk0 using the software Midjourney. With daunting synth and flute music in the background of the video, I found that the compositions were quite calming and nostalgic despite their striking gothic subjects. They resemble something I can’t quite put my finger on. An old comic book I read when I was eight? My foggy memories of The Princess Bride? But why should I be feeling nostalgia for 1970s Dungeons and Dragons-esque media that I wasn’t alive to experience in the first place? Well, the comments section had my back on this one.

Apparently it’s a real feeling. In fact, it is experienced so widely that a web page called The Dictionary of Obscure Sorrows has coined it “anemoia”: nostalgia for a time or place one has never known. At first this feels like an oxymoron. If we’re feeling nostalgia, which implies a remembrance of the past, then how can we find it simultaneously unknown? I’ve theorized that this combination is created due to the software’s blind spots. AI is still learning how to make believable images by analyzing photographs and other artwork. So, it is very possible that there are gaps in its knowledge of our world. To accommodate for those blind spots, it has to assume certain aspects of a composition—which can inadvertently create moments in an image that do not resemble anything in real life. AI is combining the known art of the past (in this case, 70s and 80s film and comics) with unforeseen imaginings of its own. This would explain how torn I am between the comfort and simultaneous mystery of these images.

These next two creations uploaded by shuj1nk0 and tenthousandscrolls seem to have the same effect. The software takes advantage of our knowledge of early comic book styles and other dated media (hence, the graininess and soft shadowing consistent with printed works at the time) which makes us feel familiar with what we’re viewing. But the subject matter itself is more difficult to place—what exactly are these structures? Does architecture like that actually exist?

We often don’t have these same questions in response to art made by humans. In Bladerunner (1982), for example, set designers created a false version of Los Angeles using pre-existing elements of the human race; the language spoken is a Chinese-German-English hybrid and the architectural scene is an amalgamation of American and East Asian styles. So, even though the environment created is entirely fake, it is wholly rooted in motifs the world is already familiar with. This is where AI makes a fundamental shift, as we can never be confident that its creations fully stem from the human world and sentient knowledge.

The curiosity that these images provoke aren’t nagging enough to cause discomfort, but do make us aware of AI’s ability to curate distantly familiar worlds—a perfect recipe for anemoia. Similarly to common themes in gothic literature or horror films, I would suggest that anemoia can be categorized as a type of uncanny, which Freud defined as a “class of the frightening which leads back to what is known of old and long familiar.” By his definition, anemoia should be a terrifying, unsettling feeling. But the uncanny in the context of AI art is not interpreted that way; instead, viewers are captivated. A user comments: “I love how these feel unnatural.” Another: “the fact that AI art has some uncanny valley elements actually makes this way better.” Commenters are craving more forms of this unsettling media, even asking for book or movie recommendations with this indefinable vibe. Our collective anemoia needs to be resolved somehow, and people are on a mad search to find a way to do so.

I, too, am captivated. AI has given me a taste of something unique and otherworldly that is hard to come by in human art. Dare I say…it’s curing my boredom. It can be tiring to view art that always originates from repeated subject matter. This is precisely why new artistic movements have consistently formed over time; as the pool of art at any given time becomes saturated with similar styles and subjects, artists will eventually find a way to invent something new. Midjourney, unknowingly, is doing just that. AI art is breaking through monotony by adopting a complex method of creation that is inherently inhuman and difficult to replicate. No wonder there’s layers of mystery packed into each one of its concoctions.

So, what does all of this implicate for the “good” art debate? I know I’m entranced by these AI works, but that doesn’t necessarily make it good. However, we do know that a useful indicator of quality art is its ability to reflect the current condition of modern life. In the early 20th century, Kazimir Malevich’s Black Square—despite being literally just a black square—was praised due to its apt representation of modern logic; more specifically, the novel idea that human expression could be depicted without imitating natural forms. But, as modern ideas change, so do the subjects and ideas that artists find important. And today, whether we like it or not, AI is permeating numerous aspects of daily life. To accurately depict modern life in our day is to embrace and acknowledge the presence of these technologies.

Regardless of this reasoning, there will always be some TikTok comments to provide opposition. On shuj1nk0’s post, cult_of_personality claims they’re “gonna cry” if the images are AI, and gabe “feel[s] like a failure of an artist” in comparison. I certainly understand their grievances. After all, art is meant to be the ultimate expression of humanity’s inner self, not a computer’s, right? But if we instead choose to value the emotion that the viewer draws from a work, the identity of its creator becomes less crucial in our evaluation of its impact. It is clear that artificially generated images have the ability to elicit human emotion—why should those emotions be any less valued than those drawn from manmade origins?

Based solely on the number of people engaging with these images and the tenacity with which they are doing so, it’s apparent that its artificial origin is not detracting from the art’s overall effect. Additionally, to be able to fabricate a shared feeling of anemoia among thousands of viewers while simultaneously converting the uncanny into a desirable aesthetic is a true feat. AI is tapping into a part of our senses that we barely knew existed and forcing us to confront our connection to both art of the past and art of the unknown digital future.

So yes, it’s good. Scary good.

Do you enjoy reading the Nass?

Please consider donating a small amount to help support independent journalism at Princeton and whitelist our site.