A Journey into How Artificial Intelligence Sees Adventism
Most Adventists describe artificial intelligence within a predictable spectrum. On one end, “How do we use this new capacity for mission?” On the other, “Is this the digital tower of Babel?” Such orientations are expected for a Protestant worldview birthed from a media innovation applied to sacred scripture; our theology has been a kind of machine thinking since the printing press, but we have taken our technological codependency for granted. Our reactions to AI highlight the need for a more robust media theory to inform our theology.
However, these are not the only ways to view such displays of machine learning. Artist and author K. Allado-McDowell has written extensively on the artistic and spiritual implications of AI. In a recent article, they write, “It’s important to think critically about how AI generators augment, amplify and ultimately colonize human imagination. . . . In the 21st century, art will not be the exclusive domain of humans or machines but a practice of weaving together different forms of intelligence. Like any relational practice, communication produces ideas unthinkable by single individuals. When humans develop deeply responsive practices with AI, we are able to think beyond our own scope. Layered human-machine collaboration produces outputs that are always both human and posthuman.”
In the spirit of Allado-McDowell’s work, I would like to engage with AI art, not as the future of art but as an example of this cyborg artistic imagination that could produce a literal image of our faith. What happens when we input “Seventh-day Adventism” into an art generator like Stable Diffusion? And what might that show us?
As far as I understand, AI art generators like Stable Diffusion operate by generating an image of noise and reiterating the process of recognizing patterns that match the text input provided by the user. It is the digital version of finding the sculpture in the block of marble. Stable Diffusion also operates off of models and checkpoints that are trained on large datasets of images. There are more and more models and checkpoints available every week, but for this experiment, I used the basic model that most tutorials recommend (v1-5-pruned-emaonly.ckpt). I don’t know how that checkpoint was created, but I am operating on the assumption that it was trained with a wide swath of images from the internet.
In generating AI art from the input “Seventh-day Adventism,” we are effectively looking at a cyborg imagination representing our faith. Analyzing these images reveals our own imaginations, not unlike a Rorschach test, as these images aren’t discrete representations but leave a large space where different interpretations can be made.
I invite you to look at these images and describe what you see to yourself. Then, read my description, and notice your thoughts or feelings that arise in response to my own interpretation. Remember, this is an exercise of imagination.
The first image appears to be in the format of a meme. Compared to the other outputs, it could also be representing a book cover within the square format (512×512 are the default dimensions). The image seems to represent robed disciples and heavenly angels oriented around the solar rays emanating from—I don’t know what to call it. A faceless face? An embryo? All of the faces are quite difficult to comprehend, almost impressionistic in style. The colors beyond the rays and bodies of disciples and angels appear verdant and bring to mind how trees in Eden may have billowed. The disciples seem to be dressed in red and blue, with some yellow and a couple of shades of green. Their bodies are ambiguous but generally evoke a posture of supplication.
The second image looks like an illustrated pamphlet or publication. In the blue field, we see a bisected circle that almost looks like a representation of the two tablets of the Ten Commandments. Instead of commandments, we have four representations: On the left side, we see a Y, the base of which looks like a tree growing out of some shrubbery as if it's a medieval illumination. Underneath the Y, we see a letter that looks like a Greek lambda. On the right side, we see a brown circle—to my eye, it looks like a sack of gold. Underneath that is an I that seems to be amidst verdant growth. The space between these two sides seems to be a brown axis that looks like a tree from a medieval illustration. On each side of this circle are flowy figures that seem like angels. Underneath the blue field seems to be some text on a yellowed “page.”
The third image seems to be the simplest. I initially saw it as a praise album due to the square dimensions, but the texturing of the brown almost seems like a weathered leather Bible. CD or book, this does seem to be a cover without any images, but the textured background does evoke a sense of sunlight descending onto the earth.
The fourth image seems to be an intricate mural of Adventist tract and book cover illustrations. Underneath a green banner, we see five columns. In the first column are four images. The topmost I cannot discern—perhaps a transmogrifying personage kneeling in a green field? Underneath we see a landscape and the first of several figures in white robes. To my mind, these white-robed figures seem to be either a depiction of a heavenly Christ or perhaps an image of the papacy, two opposites in historical Adventist theology. In the second column, we see the numeral "1," a blue-green angel facing downward from a fiery sky, a green chevron pointing downward, and a black field of swirling white “brushstrokes.” To me, this portrays an anxiety of a Sunday law and a fallen angel.
The third column contains an orange tree, a long-necked figure standing in a white dress facing away from us and toward a green canopy, and a figure in white robes with outstretched arms underneath a purple sky. Compared to the other figures in the white robes, this one seems to have some sort of blurry implements in their hands and on their shoulders—there is almost a shamanistic quality in this posture. In the fourth column, a more-than-human character in a white robe positioned between scarlet pillars stands above another figure who is chest-deep in a downward triangle of white light and has scarlet wisps for arms. Underneath the golden triangle is a black field, where a rotund figure stands, looking to me like the figure of the pope. The last column almost seems like one cohesive image: A golden obelisk stands shining toward heaven in a field of wheat. Three or four figures look up across a pool to the obelisk, and behind them, one figure in scarlet with a pale bald head looks in the same direction, standing beside a bale of wheat.
About a year ago, artist Steph Swanson, known as Supercomposite, discovered what the internet called the first demon of the latent space. She did this by entering a negative prompt, which asked the AI model to create the opposite of the starting term. Then she transcribed the words in the generated image and fed those words back as a prompt. Images of the disturbing generated character went viral.
I tried Supercomposite's process, offering “seventh day adventist” as a negative prompt, to see what Stable Diffusion views as the demon of Adventism. The results were confusing in their mundanity:
On one hand, this minimal negative prompt of “seventh day adventism” might also reveal more about the model/checkpoint and the kinds of captcha images it was trained on. Adventism is innocuous enough to reveal the machine at work, uninterrupted. But on the other hand, it’s meaningful and funny! I don’t know if it’s just the SoCal Adventist communities I grew up in, but each of these images calls to mind particular individuals from a particularly Adventist family. If I were to guess what the prompt was from these images, I would have said something like “suburban cheugy wealth.” The irony this input-output evokes is one that is deeply encoded into American Christianity: the anxiety of material possession. Are we possessed by our possessions?
Out of morbid curiosity and a desire to find more interesting images, I attempted the same with the “words” in our initial Adventist images. I did my best to transcribe these letters (SEVIHCH D DAYY SCHIARDD Siwth Toath 9AmastNothnAty RATTIN ATH DANDY. DUDIODAIIAN. ATTU MTTIM SOMIDOYY erth/ Aventad Avennt SAnl) and put them back into Stable Diffusion to see what it would give me. Googling these strings of letters gave me results pertaining to D-Day and Normandy. While the generated images seem to reflect a military wardrobe and an early 20th-century style of portrait photography, there is a distinctly Arab and Middle Eastern quality to the people (and puppets?) generated in these images.
In their article about what AI images mean for art, Allado-McDowell is skeptical and cautious about claims of digital demons but says that they point to the desire to meet entities through neural-net space. Through an artistic engagement with the cyborg imagination portrayed through AI art generators, what might we find? The answer is clear: it depends. What are we looking for? Are we hoping to replicate the style of Harry Anderson? Or the covers of unpublished Adventist books and tracts? Or the warriors those book titles seem to be illustrating?
I call these AI-generated images art even without a human hand crafting them. It is a new medium of art that will impact our imagination like every other media technology has. I felt deeply disturbed when I had my first dream of using AI art generators, possibly as disturbed as the first time a Phoenician dreamed about cuneiform. The capacity to see the art of this technology lies within our own imagination.
Philosopher Bayo Akomolafe writes that “the idea that humans have ‘natural intelligence’ . . . [is] more consequential, more interesting, than forecasts about an impending AI totalitarian takeover.” Akomolafe considers that perhaps all intelligence involves varying degrees of artifice. He continues, “Perhaps one thing to do is to trouble the assumption of distance between the artificial and the natural.” Perhaps AI is here to teach us something new about being a human, about being incarnate, about the life and agency of the images within our imagination. What AI seems to threaten is the human exceptionalism inherent in our Protestant theology and materialist science. But what threat does it really pose to us, whose Book has reached into our dreams and those of our ancestors? To us, the inheritors of living words?
To end my first Adventist experiments with Stable Diffusion, I copied the opening paragraphs of Ellen White’s first vision as the prompt. In it, perhaps we may find some hope for the future of our history, of prophecy, and our own limitless imagination.
Daniel Verdugo is a writer who grew up in the Southeastern California Conference and writes about the Adventist faith and imagination in his weekly Substack, Post-Apocalypticism.
We invite you to join our community through conversation by commenting below. We ask that you engage in courteous and respectful discourse. You can view our full commenting policy by clicking here.
