What’s your background and how did you get into AI art?
I am a writer who sometimes thinks in pictures. Or something like that. I always had a contention, where I wanted to write but also work visually. I spent a lot of time making graphic novels and filmed video sketches, but somehow I didn’t feel the mediums were aligned with the times. At university, I studied semiotics which eventually led me into VR production. There I worked a creative role for a VR/AI startup.
Adding to that, I also publish personal articles about how society reacts to technology with funny undertones. That’s what led me to text-to-image models. I wanted the AI to illustrate silly but appropriate thumbnails, quickly and without expectations. That’s how I got severely sidetracked into what’s today. It’s probably in the top 5 best decisions I’ve made.
What drives you to create?
It’s a place to put all that paradox, weirdness, and disappointments I have, but would often be too bizarre in conventional contexts. Art is what would make imagination constructive instead of destructive. And it’s good to point out that very few control what they imagine, they just do.
Anyway. There I can contend with hyperbolic ideals that need to synchronize with reality or, sometimes, the other way around. It’s really about working out inner and outer conflict but more as an investigation of reality, rather than therapy. I find it strange that things operate within the same universe yet have wildly different desires, structures, elements, or what-have-you.
But at the most basic level, because it rewards like nothing else. Creativity sparks euphoria and euphoria sparks creativity. Pleasure is pretty nice, yeah.
What does your workflow look like?
Stable Diffusion. It is the text-to-image model with the highest degrees of freedom, at the ironic cost of unstable results. In a nutshell, it’s about finding a sequence of unique prompts with “common denominators”. The commonalities would typically be visual, whether in shape, color, or movement. Once a foundation is formed through this denomination, it becomes easier to add more. One way of exemplifying this would be to combine
white sheets and
a bucket of people flows into a swimming pool where the flow or movement of the words would be one of the denominations that create this latent world. To me, this process is something in between semiotics and mathematics. I put heavy emphasis on the prompt and let the AI guide me where I need to be.
I am also meticulous about prompt weighing and use negative prompts sparingly. Img2img is a great way to direct composition and colors. In-painting is the most useful tool of all. With it, you can combine several prompt sequences into one image, work on others’ artworks as a foundation, or correct some not-so-happy accidents. I always curate outputs through the lens of if I can continue the work with in-painting, often combining different prompts.
Inspiration is much more difficult to speak clearly about. Sometimes it is about reacting to what the AI puts out. Sometimes it’s about thinking up something clever that’ll make the AI understand my vision. But all in all, it’s about finding something meaningful.
How do you imagine AI (art) will be impacting society in the near future?
Achieve more with less. Small studios’ ambitions can grow and the creator economy will continue to niche away, fragmenting cultures into the smallest size it can. What Marvel Studios can realize Today might be the work of a handful of people within decades. Fantasies will be overwhelmingly pervasive, especially when AR comes around.
In my most optimistic outlook, it will take a turn towards a more “open-sourced” or “participatory” world, with such a low barrier to entry. Whichever digital platform(s) can make fun collaborative games or worlds out of AI-generated content will have the next generation. It could pave the way for something way more meaningful, where we find a better relationship between people to people and people to machines.
At worst, it’ll become what you see when you visit #aiart on Instagram, but in immersive 3D. An overflow of plastic and sexualized content will drown out anything with actual meaning. Gatekeeping, then, might be in demand.
Who is your favourite artist?
Within the AI art category, three artists convinced me that this AI thing was worth pursuing: Theo Chronis, Ganbrood, and Phosphor. To me, they defined and embraced what was unique about the AI medium.
As for painting I mostly enjoy looking at religious, occult, and mythological motifs.
Maximalist writers such as Thomas Pynchon, Roberto Bolaño, and David Foster Wallace have found ways to contrast themes in an intense polyphonic fashion.
Seemingly, my taste in music is in a similar vein: heavy ritualistic metal, polyphonic harmonies, and layered electronic music. I don’t mind chaos.
Finally, I’d like to mention Robinson St. George who also was included in the AI Surrealism exhibition. His art seems to me to delve into an inexplicable mystery of shapes and memories. There are only questions to be asked there, which I truly enjoy.
What is your favourite prompt when creating art?
A battle of... will get you interesting results as the AI confuses contexts. In the series, Participatory Tennis is where I realized it, and so tennis was made into the ultimate democratic game:
Here is another type of battle where I did my best to blur the lines between the different characters:
It’s good to point out that the more diverse examples from history the AI has been trained on, the funnier results you can get, and the more you can combine it.
Do you have a specific project you’re currently working on? What is it?
My first graphic novel illustrated by Stable Diffusion is what occupies me now. It’s about history and time, and a bunch of other odd things. It’s one of the most challenging things I’ve done so far. But let’s keep it a secret until it’s out. Maybe you’ll like it.
Would you tell us about the AI Surrealism exhibition in NYC?
When you work with AI-generated output long enough, you become talented at distilling enormous amounts of images. That’s what the curators managed to do here, and it’s amazing! It’s no easy task to find commonality and the right framing with 100 different artists. But through this exhibition, I realized in part what AI Surrealism means. If you don’t know, you need to view the exhibition now.
In my experience the whole process was smooth. The responsibilities were expertly handled by the Exquisite Workers curators and the Superchief Gallery NFT. My two pieces and I were really strapped in for a good ride. I wish I had the opportunity to be in NYC to give a full review. From the photos it looked like a dynamic experience.
The first piece Aevum, shows this entity soaring over a blue sky. The title is taken from a medieval concept meaning a state that’s between eternity and temporality. Back then they attributed to angels who were deemed to have a changeless, constant spirit yet would appear throughout the shifting spectacles of time.
In my view, paradox, which is just another way to say the mystery of logic and experience, is essential for art. The Danish philosopher Kierkegaard said paradox to be the passion of thought and I feel that strongly. It’s only art and dreams that can reconcile impossibilities and give them a very truthful experiential dimension. Investigations into the mysteries have the potential to reveal that which can erode all previous understanding.
So the underlying idea was to reconcile impossible chaos with coherency. I went back and forth for months, and Stable Diffusion certainly gives results in either directions but to balance them is the actual challenge. In the creation of Aevum, I was experimenting with different ways to “trick” the AI by feeding it images that included warped and contradictory perspectives. I imagined this figure to be embedded and folded into spacetime, one-half confined to the linear materialist view, and the other to the quantum superposition.
With A Child’s Worst Maelstrom I sought to reconcile evil or darkness with a vibrant and playful outlook, through experimentation with colors as the init image.
The idea came to me through the children’s cartoon Adventure Time. Those that have watched it are generally struck by its depth while still upholding the classic cartoon ethos. This juxtaposition reveals how peculiar the most elementary, childish and playful things can be. So I tried to approach this juxtaposition from a slightly different position, where evil would be given a playful face and bounce the different qualities of each other.
What does it mean to be an AI Surrealist for you in the times we live in?
One telling thing I noticed when listening to the other artists speak was that few considered themselves pure surrealists. They would rather say they use surreal elements in their art. The trend nowaday is about channeling our unique selves, with our eclectic array of preferences, instead of being a manifestation of an ism. So in all, it has become quite hard for many to accept the surrealist title as it would feel like a limitation to their persona. It’s also fun, that with AI you can channel the different movements of art history with ease, so the many ideas can be used simultaneously as they’d be palettes of paint.
However, surreal AI art has some interesting things to think about, that differs from the classical sense. It can be hard to argue that AI art stems only from your unconscious dream associations, which would be the essence of classical surrealism. In prompting you work with conventional understandings of a keyword, that is the internet’s understanding combined plus complex mathematics. So no matter what the AI dreams up it’ll always be an averaged dream originated from people of different parts of time and space.
In conventional language many would say that the average would be something very normal, but many of our AI surrealists have really pointed out how bizarre the average can be. The art of Roope Rainisto is one of the clearest examples. The latent space that AI creates is different from humans’ understanding. The latent isn’t shaped by eons of knowing what it feels like to bounce about in Euclidean space. The AI often outputs something that feels like the thing we referenced, but also lacks something essential. The missing essence comes from the AI’s lack of lived experience, and gives it the surreal tone.
It feels we’re on the precipice of something here, where living through data will come overthrow all previously held beliefs. Quantification will be regarded as the undoubtedly true way. Subjective experience in the coming era will be kept in the dark, as subjectivity can only reach its potential as a quantity in a larger database. This anticipation of mine, creates a strange relationship with these machines. The AI surrealist can perhaps feel this and instinctively expresses it.
Anything else you would like to share?
As we get more efficient with these machines and the increasing output it’s important to spend more time introspecting and synthesising. Is the same idea with social media where catch ourselves in the scrolling, and the consumption of new information wouldn’t really lead us to what’s meaningful. Quality over quantity, is what I think the world wants right now. Or well, at least I do.