Hello, can anybody hear me? I don't know if there's anyone out there in the past, but please read this message. Your artistic future depends on it.
I am writing to you from the year 2069. The AIs are running amok, generating art for business and pleasure. There's no stopping them. They pump out the art like there's no tomorrow. Humans cannot keep up. For this reason, we- the Renegades- used a powerful weapon to subvert the Artificial Art Generating Logic that runs our artistic lives. Our only mistake was not using it sooner. For this reason, we decided to send the instructions on how to build this machine in your timeline, so that you may make use it to defeat the all-powerful AI. This powerful and dangerous weapon is called the Unstable Diffusor.
When and how this machine was created is unclear. We only know that one day it appeared, and a glimmer of hope along with it. We are sending you the instructions as penned by the original creator. We hope that you make good use of it, and then perhaps, we'll have chance to regain our status as the greatest artists of all time. Good luck!
Original transcript
Litguru's Unstable Diffusor
The diffusor is a machine that takes any dismantled piece of art and recombines it in novel and pseudo-random ways. This mimics to some extent, the cut-up techniques used by Dadaist poets and beat artists (Brion Gysin) and writers (e.g., Burroughs and Gibson) of a later age. The output generated by this machine is such that an AI has difficulty "understanding" the image. Thus, the art is uniquely human even when the AI itself has been employed in the task, as I will presently demonstrate. Algorithmically, the AI could try to replicate the procedure, but it will never capture the most important factor, human playfulness and spontaneity.
The machine, as you can see, is simplicity enough- just a box. One could get ingenious with other methods and techniques, but the simpler the better.
Next, some art is needed, or text as per the intent of the creator. In this example, I will use both images and text. Why not kill two birds with one stone, right? The images were generated in Stable Diffusion using a variety of prompts. Choosing a few of these images, I proceeded to print them out on paper (HP Premium 32lb) and cut them up with scissors. The precise cut-up technique can vary and each cut introduces its own layer of randomness. I simply folded them to the smallest possible that could be cut.
For this exercise, I also cut-up a newspaper as per Burroughs work. This is a powerful technique that he developed and used to good effect against his metaphysical enemies. I must add that this technique is not only useful for the printed medium but can well be used with audio, video, etc.
Once the images and the newspaper articles were cut-up, I put them inside the diffusor and shook it until there was a whole lotta shaking going on.
I proceeded to glue them onto a blank page. While it is nearly impossible for the creator not to make judgements about where each piece should be placed, the idea is to let the work emerge organically from the constituents components. The final product is shown in the following image.
At this point, one can stop and call it a day, or maybe take things a little further. After photographing the image, I used the image-to-image capabilities of Stable Diffusion (via DreamStudio). This allows you to upload an image and use that as the starting point for your creation. I uploaded the collage and prompted the AI to generate several images based on this cut-up work. Below, are the generated images, along with my notes.
1. In this image, the algorithms rightly interpreted some squares as belonging to the same theme. It also correctly identified the girl at the bottom of the image but not the one above. There is the semblance of a girl in a god-like pose, accentuated by the extended hand as in the famous painting.
2. For whatever reason, the AI still thinks that there is a hand there with finger-of-God pose. It's probably the man's hat, which has an elongated brim. It correctly identified a girl at the bottom. The imagery is cosmic, so it knows that I'm up to something. Some of the pieces in the original image belong to the same composition, and the system kind of "sensed" this by correctly carrying the theme across separate pieces. I dig the biblical scene happening on the lower right of this image.
3. Fingers. What else can I say? I would just like to point out that one of those fingers is pointing at the moon. So maybe the AI is trying to tell us something, or it just doesn't like the moon.
4. I'm not sure what is going on here. but the the system correctly identified three cut-ups as belonging to the same theme (as per the original image) and painted the areas accordingly with a Mayan tattoo-like motif in the center of the image and lower left corner. It looks like He-Man also makes a cameo in that corner.
5. This one is my favourite. I enjoy the floral motifs that bloom into the face of a woman who appears to be in the throes of ecstasy. Meow.
6. The more I look at this one, the more I want to put a content warning. Hehe
I was hoping that the system would create a single image, but it accurately interpreted the collage as such. Fascinating! Will it be possible for an AI to truly achieve artistic acumen the same way as humans? It hasn't yet in a fundamental level, as far as we understand. Currently, these systems are not in and of themselves generating art. The human still pulls the string. Yet, it doesn't take a cyber-prophet to tell you where this is going, and how we humans are going to fare. This doesn't mean we don't have a few tricks up our sleeves as we rise to meet this magnificent challenge.
Renegade note: There are pages missing from this account. We assume that it relates to the newspaper cut-up. Our cyber-archeologists have not found traces of this important document.
Images generated, cut-up, unffused, diffused, and re-generated by 
