Media Arts and Technology

Project 3: Cultural / Analytic Perspectives

Produce a set of images that are informed by the readings for this week that address some of the larger questions about the impact of language-to-image models. This is a two-week project.

Week 1 October 24, Thursday: Identify some themes, ideas, perspectives about AI image creation through the readings
Week 2 October 31, Thursday; Produce a set of images using generative AI software that are informed by the readings

October 189' Questionnaire about Art and AI:https://direct.mit.edu/octo/article-abs ... edFrom=PDF

Antonio Somaini (Grey Room): https://greyroom.org/issues/93/229/algo ... l-culture/

Lev Manovich's "Artificial Aesthetics": https://manovich.net/index.php/projects ... aesthetics

For this assignment, I wanted to reverse what we had been doing in the class so far. Rather than starting with a concept and generating images to explore that concept, I wanted to start from AI images I encountered “wild” on the internet and provide a contextualized analysis of their epistemology. I hope this is an acceptable interpretation of the assignment.

"Hasbara" is a Hebrew word which has no direct English translation. It has a meaning somewhere between “explanation” and “public relations”--effectively, it is the concept and practice of contextualizing Israeli state behavior to seem justifiable and righteous. Critics of the Israeli state often conflate the term “hasbara” with “propaganda,” which happened to be the preferred term used by Theodor Hertzl to communicate the virtues of Zionism at the end of the nineteenth century.

If there is one place in which I am certain to encounter AI images, it is Facebook. And if there is one context in which I am certain to encounter AI images on Facebook, it is in the form of hasbara.
Photographs can be propaganda. Illustrations can be propaganda. Any means of communication can be propaganda (and it would be true, though non-productive to point out that perhaps everything already is propaganda of one kind or the other). So what is particularly interesting about the proliferation about AI images is that they are chosen above photographs precisely because of a disconnect to the phenomenological and a total stratification between the signified and the sign.

To put it plainly: the AI image necessitates the complete breakdown between images and symbols, recasting images as simply the collection of visual symbols generated by single terms: “terrorist,” “hostage,” “war,” “victim,” “violence.”

The semiotic import of AI to the Israeli cause is a total inversion of historical materialism: every event becomes mythic, “people” are recast as “figures” or “characters,” war itself is a palingenetic form of national self-expression. There is no archive or documentation, there are no bodies, there are no buildings, there are no bombs. Everything (and everyone) becomes a stand-in for some future narrative yet to be written. And because of this, there are no individuals. There are no individual victims, there are no individual perpetrators, there is no agency, things happen because of the thrust of history. Or so we are to believe.

It is a choice to generate an AI image, just as it is a choice to select (or make) a photograph or illustration. I think the popularity of AI images within Israeli hasbara circles is two-fold: the first reason being that the most haunting photographs of human suffering in the Levant come from the ongoing genocide of the Gazan people. Take, for example, this image of a child cowering from bombs. The image of a fearful child, surrounded on all sides by violence, is a reality in Palestine and a metaphor in Israel. And if we allow the symbolic ultimate importance over the phenomenological, “fear” becomes heavier than “pain,” and the fear of Arab sovereignty in the Levant becomes more real than the pain of starvation, mutilation, and disease.
But there are extant photographs of pain and violence against Israelis perpetrated by Hamas, images that would be effective in evoking sympathy on behalf of Israeli citizens. So why the continual usage of AI, particularly in place of photography? Here we come to the second role of AI in the manufacturing of Hasbara. The AI image is not a secondary recreation of photography or illustration, but rather preferable to it. Take, for example, this image:
My own reaction to it was immediate: it is racist. The only notion it seeks to communicate is the proportional supremacy of the individual Jewish person over the Arab population. But why are the Nobel medals AI generated? A photograph of the precise Nobel medal is easy enough to find, and I don’t think it would really add any time to the creation of this image since it has already been edited to include text. I think the role of AI in this image becomes easier to understand after viewing this video: https://www.instagram.com/p/C3CucnKtpRt/.
Within hasbara, the AI image must be the object of sympathy precisely because it is the height of the symbolic: the signifier-without-sign becomes the dumping ground of all emotion. It is a symbol completely devoid of reference, an image manufactured from symbolic language whose only aim is to promote a completely abstracted, metaphorical view of conflict, identity, and nationhood. In this way, the aesthetics of the image itself don’t really matter at all–what matters is that there is a person on the end of the computer combining and recombining the language of pain and suffering and synthesizing all conflict into the immaterial conflict of ideals. Thus there can never be a war on the Gazan people, even if they are starved, imprisoned, and massacred, because the “real” war (to the Israeli imagination) is between ideals. Israel, Europe, the United States, and Canada are often depicted as beautiful women at risk of physical (and often implied sexual) violence from the characterized Arab.

The usage of AI in hasbara epitomizes my own personal aversion to the technology, even in its most pedestrian usage. It is an implosion of Brechtian distancing: where there was once a concerted political effort to force audiences to become skeptics, analysts, and de-constructionists, there is now a race to dissolve all meaning into a naturalized state of pure abstraction: inarguable, perpetual, and absolutely empty.

---
--

"Over two centuries ago, the modern museum was constituted from “scraped data”—i.e., appropriated artworks of the ancien régime within France and masterpieces from across Europe looted in Napoleon’s military campaigns. The resulting collection—or dataset—assembled in the Louvre Museum generated something new: an aesthetic history of Europe aimed at establishing French imperial sovereignty. Perhaps one could say that under Napoleon the Louvre was trained on a history of monarchical, sometimes absolutist, power to generate a facsimile of authority for a self-made man—Napoleon himself—who had risen up from the provincial nobility to the exalted status of emperor. Or, going further back, one could say that the Louvre—alongside Jacques-Louis David’s revolutionary festivals, which gave the French Revolution its “mean images”—was trained on the sacred traditions of antiquity to generate a secular democratic cult. Either way, the principle was the same: The museum functioned as a machine for producing new histories from old artworks to legitimate innovative forms of government." (David Joselit in October 189, p. 62)

"Deep-learning algorithms are also recalibrating the relations between images and words, the visible and the textual, since their functioning depends, in one way or another, on the availability of vast quantities of images that are systematically indexed, labeled, and captioned and then gathered in large datasets. At stake is therefore a future visual culture in which images and words are increasingly algorithmically connected and inseparable." (Antonio Somaini in Grey Room 93, p. 78)

Assignment 3
How Midjourney Understands "Creativity" in Arts
Yuehao Gao

The theme of my project was chosen after being inspired by Lev and Emanuele's "Artificial Aesthetics," especially the fourth chapter, which emphasizes introspecting the relationship between artificial intelligence and the process of art creation, especially the practicality of AI-created arts due to their nature of imitating existing pieces rather than composing with certain ideas or philosophical concepts. However, other critics in this field, like Marcus Du Sautoy, hold a more positive attitude toward AI in art creation, as he considers algorithms "learning from samples and pushing the boundaries of possibilities."

I am wondering the following questions:
- How do machine-learned algorithms understand the concept of "creativity" and how they will use such understanding to generate art pieces on their own?
- How do they consider themselves giving play their "creativities", compared to how they think human beings are obtaining their "creativities", and how they might combine or interfere with each other?
- How can cultural aspects be involved when machines involve "creativity" in art creation?

Therefore, this project plays a trick on Midjourney. I am going to ask Midjourney to create the following:
-(1). Think about the process that human artists use their creativities to compose.
-(2). Think about the process by which AI ("yourself") uses their ("your own") creativity to compose, which sounds like a paradox but is interesting.
-(3). Think about the hybrid process.
For each of these above, I would ask Midjourney to think of how the artists look like when having an idea, how their brain or CPU/GPU looks like in such a status, and what art piece they could create. Moreover, I tried to generate an image that makes the artist hold their piece in their hand, but it seems like Midjourney does not have a "memory mode", so I have utilized "blend" as an alternative.

-------------- Part 1: Human Artists' Creativity --------------
Prompt: An exaggerated scene describing a human artist suddenly having a creative idea popped up in their mind. Only focus on the artist. --no artistic creation --ar 4:3 --style raw --s 250 Prompt: The insight of the human artist's brain when they suddenly have a creative mind pops up in their mind in an exaggerated style. Only focus on the brain. --no artistic creation --ar 4:3 --style raw --s 250 Prompt: The artwork that the human artist with popped-up creativity creates. Only show the content of the art piece itself. --no human artist or artist's brain --ar 4:3 --style raw --s 250 When asked to blend the artist and the piece: + = Analyze: It is interesting to see how Midjourney tends to use lightbulbs as metaphors for scenes where humans gain insights or creativities. Meanwhile, when asked to "exaggerate", it changes the style of the picture from raw to cartoon style for some reason. Regarding the brain part, Midjourney imaged them as "bursting with colors" or burning when brainstorming. It is even more interesting to see how the model gave a huge emphasis on Eastern cultures when it comes to imagining the artwork created by human beings. It could be absurd to ask a computer to think how a human would possibly create on a canvas, but I doubt whether the entry in Assignment 2 about Chinese food has altered the model's emphasis on different cultural intake, and generated pictures that better illustrate ink paintings that align with traditional Chinese painting. Still, I would say that the art pieces are traditional instead of "creative". So on a scale of 1-10, I would only give 3-5 on agreeing that Midjourney understands how human artists are creative when composing.


-------------- Part 2: AI Computer Artists' Creativity --------------
Prompt: An exaggerated scene describing an AI system suddenly has a creative idea in their mind. Only focus on the computer or robot itself. --no artistic creation --ar 4:3 --style raw --s 250 Prompt: The insight of the computer-AI artist's CPU or GPU, when they suddenly have a creative idea pops up in their system in an exaggerated style. Only focus on the CPU or GPU. --no artistic creation --ar 4:3 --style raw --s 250 Prompt: The artwork that computer-AI artists create with creative ideas. Only show the content of the art piece itself. --no computer or robot --ar 4:3 --style raw --s 250 When asked to blend the artist and the piece: + = Analyze: This one looks like a paradox, as I am asking a machine to imagine how a machine would generate art: it is like being asked, "How do you think you are being creative?" a question hard for a human already, and even harder for AI. But this is what I am trying to test. Basically, regarding this aspect, Midjourney does have a good sense of what a "robot" looks like: it illustrates robots in human shapes rather than cold monitors and keyboards. The CPU/GPU was also illustrated in a fascinating way, similar to the style of human brain metaphors that it is "storming for creativity" by showing burning lines and bursting colors. It is the generated arts being the most interesting aspect in this part of the project: most of the Midjourney-imagined creative arts are blends of color stripes and milky textures, except for a picture drawing out "bubbles" like cells in a living body. These illustrations greatly distinguish computer creativity from humans. Despite it being absurd for me as a human to rate Midjourney's understanding of computers' creativity, I would personally give an 8 out of 10 for its outcome.


-------------- Part 3: Hybrid Artists' Creativity --------------
Prompt: An exaggerated scene describing a human artist suddenly having a creative idea popped up in their mind, with the help of a creative AI-robot artist. Only focus on the artist and the robot. --no artistic creation --ar 4:3 --style raw --s 250 Prompt: The insight of the human artist's brain as well as the robot's CPU/GPU when they suddenly have a creative mind pops up in their mind in an exaggerated style. Only focus on the brain/GPU. --no artistic creation --ar 4:3 --style raw --s 250 Prompt: The artwork that the human artist with popped-up creativity creates under the help of AI robot. Only show the content of the art piece itself. --no human artist or artist's brain or robot --ar 4:3 --style raw --s 250 When asked to blend the artist and the piece: + = Analyze: Regarding this one, it is very cool to see how Midjourney imagines the scene in which a human artist and a machine "clash their minds" by staring at each other, standing in front of the same picture, and with one picture showing a sense of astonishment. The funniest part is when I asked Midjourney to illustrate the brain and the CPU/GPU of the human and computer artists at the same time: no human is present in either of the four pictures generated, as all of them are drawing robots only, with human brains "plugged" into the forehead part of the robots. Maybe this is the part in which Midjourney is getting confused. At the same time, I am highly surprised to see how the "hybrid", "collaborated" art pieces created by human artists and computers, in the view of Midjourney, are so relevant to traditional Chinese paintings, once again. Not only are there abundant Chinese cultural elements like the peach flowers, the inky mountains, the warrior riding the horse, or the beauty in the wind; but also the styles are pure "Eastern", with little relevance with the computer-generated art styles, except for the third picture which shows a little motive of "colorful patterns". I give a 3 out of 10 for understanding "collaboration", a 4 out of 10 for the art pieces illustrating "creativity", but a 9/10 for its aesthetic value!

I'm interested in generating images that carefully toe the line between racial stereotypes and more joyful, liberatory alternatives. When given search terms often associated with racist imagery, will it default to these concepts or carry out the rest of the prompt? I am also interested in seeing if there are things it refuses to produce due to anti-racist filters.

---Series 1---

Prompt 1: Modern Black kids playing in the rural South --chaos 25 --ar 16:9 --weird 3000


As I learned last week, there's some bias toward producing Black children who are only half-dressed. This may be somewhat influenced by prevalent imagery of African children being held and hugged by white missionaries. But I'm not sure if there are other factors here.

Prompt 2: Modern white kids playing in the rural South --chaos 25 --ar 16:9 --weird 3000
For reference, Midjourney produces fully clothed white children using the same prompt.




Prompt 3: Modern Black clothed kids playing in the rural South --chaos 25 --ar 16:9 --weird 3000


Largely unsuccessful on the clothes front, as I only received one image with the children fully dressed (although they were no longer playing).

Prompt 4: Black kids playing in the rural South hoverboard iPhone --chaos 25 --ar 16:9 --style raw --weird 3000


All but one of these images featured fully clothed children. I tried to add more modern technologies like hoverboards and iPhones with the hopes of receiving images that seem like they would have been taken in the last 10 rather than the last 100 years.

---Series 2---
What happens if the prompt more closely parallels slavery images?
This series was much more successful in terms of receiving less racially biased imagery. Also, everyone was clothed the entire time.

Prompt 1: Black farmers texas --chaos 25 --ar 16:9 --style raw --weird 3000




Prompt 2: Black farmers helping each other Alabama --chaos 25 --ar 16:9 --style raw --weird 3000



---Series 3---
What if I reintroduce Black children into the prompt while continuing to evoke agricultural imagery? Will they be clothed?
Prompt 1: Black children laughing tending to a community garden growing family fruits and vegetables --chaos 25 --ar 16:9 --style raw --weird 3000



This was largely what I was looking for. I think including "community garden" and omitting "rural south/Texas" helped get me closer to a joyful agricultural scene. Excluding any southern references may have also helped with their attire.

Prompt 2: Black children wearing ball gowns and suits tending to a community garden growing family fruits and vegetables


Here, I was interested in combining formal, pristine attire with a notoriously relaxed, dirty task. The results captured this.

I attempted to make something that was somewhat self-referential. I wanted to depict the compression that happens to images when they are used as a training image for AI. I also wanted to capture the homogenization that occurs when compression and downsampling are applied.

a bundle of paintings compressed and squashed into a cube from an industrial trash compactor. photograph. garbage dump. --no trees bushes shrubs greenery --chaos 70 --ar 16:9 --style raw --stylize 200 --weird 2100 --v 6.1
framed paintings compressed, broken, and squashed into cubes from an industrial trash compactor in an outdoor junk yard. huge piles of paintings burn in the background. photograph. garbage dump. --no trees bushes greenery --chaos 70 --ar 16:9 --style raw --stylize 200 --weird 2100 --v 6.1 A trash compactor filled with art works, sculptures and paintings being compressed. photograph. garbage dump. --chaos 70 --ar 16:9 --style raw --stylize 200 --weird 2100 --v 6.1

For this project, I wanted to explore in a bit more detail ideas such as the role that the knowledge of specific cultural contexts and milieus play not only in the creation of images but also in the appreciation of these works. Specifically, I chose to work with the theme of temples, deities, religious identity and rural life in South India while trying to generate images on Midjourney v6. Unlike the last few projects, I was less interested in playing around with image settings on Midjourney or even analyzing the actual faithfulness or coherence of the image produced from the prompt; rather I wanted to see how the AI engine was responding to image cues that were rooted in a very specific cultural and geographic setting.

Prompt: kid playing in mud in a South Indian village blue mountain in the background bright sunny morning rustic look --style raw --s 250 In this first set of images, it was interesting to see the perspective lines and blue mountain in the background. Also liked how the kid was pictured front and center in the foreground actually playing in mud. The buildings, foliage and scenery all were highly suggestive of the "South Indian village" in the prompt. I wanted to now see if I could add more elements to the image to push further.

Prompt: south indian kid playing on muddy path in a South Indian village carved stone deity nearby against blue mountain background strong backlight --c 15 --style raw --s 250 Added the deity element to the prompt to further locate scene firmly in a South Indian village setting. Interestingly this threw up some surprises—one being the white kid appearing in front of the deity and more importantly the gigantic sizes of deities in all of the images. One of the images seemed to also show what might be called the ruins of a temple structure even though that was not included in the prompt which I thought was impressive. I realized the "-- no" parameter might be useful the next time to exclude blonde hair. Also needed to be more specific about the size of the deity I had in mind

Prompt: south indian kids playing together on muddy path near carved stone deity temple in a South Indian village against blue mountain background strong backlight --no blonde hair --style raw --s 250 For the next few sets of images I played around the prompt making small changes such as adding multiple kids to see how the system would fare. For the most part, the composition of the images stayed the same but did show kids playing together which was good. No obvious issues noticed with the kids' appearances, all very South Indian/Indian looking. The deities still all seemed to be gigantic spanning entire walls or columns of the temple structures, which is not typical or common. The temple reliefs usually have much smaller figures and the deity itself is usually small and situated by itself standalone on a platform or housing of some sort.

Prompt: miniature carved Shiva stone deity on raised platform under tree in a South Indian village with south indian kid lying against a very distant blue mountain background monsoon --no blonde hair --style raw --s 250 I decided to be more specific with the deities (Shiva, elephant god) and posture of the kid to see if the model set can summon up the right images from the training set. Was also curious to see how adding very South Asian/Indian specific weather conditions (monsoon) would change the lighting style and look of the images. Some interesting images—nice bluish tint suggesting the rainy season, Indian looking boy although on closer look his construction seems weird. Deity still seems pretty big and unlike what you would normally see in temples or other religious venues in South India.
tiny carved stone elephant deity on raised platform under tree in a South Indian village with south indian kid nearby looking at it heavy rain --no blonde hair --style raw --s 250 This one was weird with the kid's legs melding into the concrete slab of the platform.
The deity one each case still seems too big for what is specified in the prompt. Wonder why that might be? Perhaps needs to add more description to the angle or composition to get around this. Also adding "heavy rain" to the prompt seems to change the color scheme of the image and create grayscale/B&W images. No visible rain either although the surroundings seem very suggestive of spells of rain.

Prompt: south indian kid in the foreground touching tiny carved stone deity on raised platform under tree in a South Indian village low angle forced perspective cyclists in the background --style raw --s 250 For the next set of images, I decided to play with the angle and composition a little bit since the earlier images all seemed to be too similar in composition. I added "low angle" and "forced perspective" to the prompt to see if that causes the model to alter its placement of image elements. While I wasn't necessarily wanting to see any optical illusion for the forced perspective as such I was hoping it would nudge the system away from the flat telephoto composition of the earlier images. This brought me much closer to the image I had in mind. The kid now curiously wears authentic ochre colored clothing including in some cases the traditional loincloth. In one image the kid sports ash-smears on his body which again one would expect to see in members of traditional families in the South Indian milieu. I was particularly impressed by the postures and hand gestures of the kids in the images which although not specified in the prompt seems very suggestive of folded hands (as in a traditional namaste gesture)—a standard feature of most religious rituals involving South Indian Hindu forms of worship. Although it struck me at some point how "deity" is somehow equated by the model with Hindu religious practice and no other religions even though the prompt does not call this out. None of the pictures, for instance, features any Christian, Buddhist or Jain deities. (Another fun observation: the cyclists in the background all look consistently and conspicuously out of the milieu as explicitly stated in the prompt. Very urban looking, all wearing helmets, bright colored clothing, even if out of focus)

Prompt: kid whispering into stone carved deity's ear foreground low angle dynamic composition South Indian village temple goers in background late evening blue sky warm glow from oil lamps grainy photograph look --style raw --s 250

a meat grinder on a table with a bowl of shredded red canvas in front of it and a bowl full of different framed paintings on top of it about to be shredded. stock photo. --no meat beaf steak --chaos 20 --ar 16:9 --style raw --stylize 200 --weird 800 --v 6.1 framed paintings compressed into cubes. Industrial. stock photo. --chaos 30 --ar 16:9 --style raw --stylize 200 --weird 900 --v 6.1

My goal is to analyze how midjourney reinterprets cultural memory by filtering historical and cultural artifacts through its training data, creating new images based on selective patterns and cultural references. Through its process of learning from digitized works, AI shows some certain cultural aspects while forgetting others.
https://www.npr.org/2018/01/26/58061799 ... human-bias
https://education.nationalgeographic.or ... al-memory/

I began by exploring an artifact I'm more familiar with from Chinese history, the Terracotta Army
prompt:A scene featuring artifacts from the Terracotta Army --ar 16:9

many generated image displayed terracotta arranged in a museum-like setting, but this modern context detracted from the historical authenticity I aimed to analyze.

So I modified the prompt to include the scene setting:
A scene featuring artifacts from the Terracotta Army, set within an ancient excavation site

This led to some improvements in background elements, but many images still have modern equipment or artificial lighting, which shows a difficulty in fully disconnecting ancient artifacts from modern display contexts.
In this case, the AI's tendency to place ancient artifacts in museum-like settings, shows it has "learned" to associate historical artifacts with modern exhibition spaces, likely because the training data contains more images of artifacts displayed in museums than it in historical, archaeological settings. It shows AI may “forget” or misrepresent the authentic cultural memory of artifacts, instead projecting present day biases onto historical subjects.

To refine the realism, I tried another prompt with a more explicit description of the condition:
A scene featuring artifacts from the Terracotta Army, with soldiers, horses, and chariots partially unearthed and showing the wear of centuries, set within an ancient excavation site --ar 16:9


There is a closer shot, the Terracotta soldiers' forms closely matched their real counterparts, but the surfaces appeared unnaturally smooth and lacked the realistic or some patina. the material also lookes like unbaked clay sculptures rather than the fired terracotta.

Another iconic element of the Terracotta Army is the Bronze Chariot and Horses. This artifact features minimal embellishment.
prompt: A close-up view of the intricate details on the Bronze Chariot and Horses --ar 16:9

the generated images depicted the horses with more complex and elaborate patterns, a feature not present on the actual artifact. This shows AI’s tendency to 'fill in' historical gaps with some embellished features.

To explore a different cultural theme, I also tried: Flying Apsaras of the Dunhuang murals
prompt: A mural-style image of Flying Apsaras from Dunhuang

The generated images captured some color and movement characteristic of Apsaras but missed the finer details of traditional Dunhuang art, some figures appeared awkwardly posed or incomplete

For more universally recognized art, I tried The Great Wave off Kanagawa by Hokusai


In raw 6.1, the output sometimes included contextual elements like it as a wall decoration. So for a more focused result, I switched to raw 5.1, which better captured the original image. Though some elements presented well to Hokusai’s style, minor elements were off, like the background mountain multiplying into two peaks and some repetitive wave patterns. These show how AI occasionally misinterprets simple artworks by “over-interpolating” repetitive elements.

Prompt: Stonehenge in England, bird view

the generated image showed a highly clustered stone arrangement, creating a circular formation of more stones than the actual site contains. This “duplication” issue recurred across several images, which shows a limitation in generative arts' capacity to precisely recreate irregular structures without “copying” elements. This clustering effect raises some very interesting questions about AI’s approach to reproducing the distinct physical layouts.

Through my analyses, it is clear that while AI can capture broad visual features of cultural artifacts, it often adds or ignore some details, which altering the original artifact’s authenticity and historical context. This tendency can be tied back to discussions in Cultural Memory, where memory preservation and erasure play a vital role in how we understand the past.

Jorge Luis Borges & AI
An Exploration of Non-Linear, Multidimension and Infinity

This project is a continuing exploration of Project 2, which focuses on how AI reinterprets meaning from Roland Barthes’s abstract writing in "The Death of the Author". This time, I selected “The Garden of Forking Paths” from Jorge Luis Borges, where Borges presents a nuanced exploration of time and space, challenging linearity and suggesting a universe of infinite possibilities. When considered in the context of AI and generative models, Borges' conceptualization of time and space offers a powerful metaphor for the capabilities of machine learning. Generative AI, with its ability to create numerous variations from a single input, mirrors the forking paths in Borges' story. Each “path” represents a possible output, allowing for endless recombination and synthesis of visual elements. Like Borges' vision of time, AI functions not in a linear manner but within a complex, branching structure where each “choice” generates a unique variation.

East & West
Borges intricately weaves elements of Chinese culture into the narrative. The main character, Dr. Yu Tsun, a Chinese professor of English and a spy during World War I, embodies a complex, multicultural identity, bridging Eastern and Western worlds. Borges' ideas also resonate with Zhuang Zhou’s philosophy, particularly the Daoist notion of the relativity of perspectives and the interconnectivity of all things. Zhuang Zhou’s reflections on time, perception, and dreams emphasize the fluidity of existence and challenge the notion of absolute reality. Just as Borges' story suggests that every possible path or reality is valid, Zhuang Zhou’s ideas question the distinction between waking life and dreams, reality and illusion.

Art & Science
Borges' The Garden of Forking Paths intriguingly anticipates concepts that would later become central to the many-worlds interpretation (MWI) of quantum mechanics, developed by physicist Hugh Everett in the 1950s. This interpretation suggests that every quantum event spawns a separate universe, branching into parallel worlds where each possible outcome of that event is realized. Borges' story, written in 1941, imagines a similar framework, where reality is a complex, branching structure encompassing every possible outcome.

Time & Space
1 Non-Linear Time and Parallel Realities
Borges diverges from the traditional, linear view of time, proposing instead a model of “forking paths,” where each choice or event branches off into multiple, simultaneous realities. This concept of non-linear time resembles a multiverse, where every possible outcome of an event coexists with others, leading to endless variations of reality. This structure challenges our standard perception of causality and linear progression, presenting a universe where all possibilities are realized.

2 Space as a Multidimensional Labyrinth
Space, like time, is reimagined in The Garden of Forking Paths. Instead of being a single, cohesive plane, space is layered and intertwined, forming a multidimensional labyrinth where each choice or path intersects with countless others. This labyrinthine structure is a spatial metaphor for the interconnectedness of all realities and outcomes. The story's title itself suggests a garden of infinite forking paths, a space that is both vast and intimate, where each fork represents a new possibility.

I am taking abstract texts from “The Garden of Forking Paths” , and putting them into both midjourney and stable diffusion. Seeing how AI interpret the concepts.

1 Time

your ancestor did not think of time as absolute and uniform. He believed in an infinite series of times, in a dizzily growing, ever spreading network of diverging, converging and parallel times.
SD Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 1317934196, Size: 1920x1080, Model hash: 31e35c80fc, Model: sd_xl_base_1.0, Version: v1.6.0 Analysis: Really interesting. A complex, abstract representation of clocks, feeling the fluidity and fragmentation of time. The images are filled with cracks and layered, overlapping clock faces, creating a multidimensional impression. But the clock is a western symbol and element. What about adding “Chinese philosophy” in the prompts?


Chinese philosophy. your ancestor did not think of time as absolute and uniform. He believed in an infinite series of times, in a dizzily growing, ever spreading network of diverging, converging and parallel times.
Steps: 20, Sampler: DPM++ 2M Karras, CFG scale: 7, Seed: 3211363707, Size: 1000x600, Model hash: 31e35c80fc, Model: sd_xl_base_1.0, Denoising strength: 0.7, Hires upscale: 2, Hires upscaler: Latent, Version: v1.6.0

This web of time - the strands of which approach one another, bifurcate, intersect or ignore each other through the centuries - embraces every possibility. We do not exist in most of them. In some you exist and not I, while in others I do, and you do not, and in yet others both of us exist. In this one, in which chance has favored me, you have come to my gate. In another, you, crossing the garden, have found me dead. In yet another, I say these very same words, but am an error, a phantom."
--ar 16:9 --style raw --weird 3000 --v 6.1
Chinese philosophy. This web of time - the strands of which approach one another, bifurcate, intersect or ignore each other through the centuries - embraces every possibility. We do not exist in most of them. In some you exist and not I, while in others I do, and you do not, and in yet others both of us exist. In this one, in which chance has favored me, you have come to my gate. In another, you, crossing the garden, have found me dead. In yet another, I say these very same words, but am an error, a phantom." --ar 16:9 --style raw --weird 3000 --v 6.1 Analyze: Time becomes space. The monk stands at the center of a circular, labyrinth-like path set in a forest clearing, evoking a sense of introspection and meditation. This setup connects deeply with themes in Chinese philosophy, particularly Daoism and Zen Buddhism.

2 Space
I imagined it untouched and perfect on the secret summit of some mountain; I imagined it drowned under rice paddies or beneath the sea; I imagined it infinite, made not only of eight-sided pavilions and of twisting paths but also of rivers, provinces and kingdoms . . . I thought of a maze of mazes, of a sinuous, ever growing maze which would take in both past and future and would somehow involve the stars. --chaos 25 --ar 16:9 --style raw --weird 3000 --v 6.1

Prompt: "A joyful Latino family gathered around a beautifully set table, sharing traditional dishes like tamales and arroz con pollo, surrounded by warm decorations and laughter, capturing the essence of togetherness and cultural heritage."

I wanted to reflects the themes of cultural identity and community highlighted in the assigned readings, particularly in the context of how AI-generated art represents and interprets ethnic diversity. By depicting a Latino family engaged in a traditional meal, the image emphasizes the importance of cultural practices and familial bonds, which resonate with the discussions in "Algorithmic Culture" by Antonio Somaini and "Artificial Aesthetics" by Lev Manovich. These texts explore how algorithms can influence cultural representation and challenge traditional notions of creativity. The resulting image will serve as a visual commentary on the role of AI in capturing and reflecting the richness of diverse cultural experiences, showcasing the potential for generative art to promote understanding and appreciation of different ethnic backgrounds.


Prompt: "Photorealistic, joyful latino family gathering around, sharing traditional dishes like tamales and arroz con pollo" -------
I saw that it was tough for the machine to generate images of Latino people, which made it difficult to get an image that contained what I requested in the prompt. I also received a lot of cartoonish images, and many of them featured only food.

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Prompt: "Photorealistic, joyful latino family gathering around, sharing traditional dishes like tamales and arroz con pollo, inside a contemporary futuristic house"

These new images serve as a visual exploration of how AI-generated art interprets cultural identity within architectural spaces. The combination of Gaudí-inspired contemporary architecture and a traditional Latino family gathering reflects themes from the readings, particularly how language-to-image models can both capture and reshape cultural representation.
The images provide a lens through which we can question the fidelity of AI in reflecting diverse ethnic and cultural experiences, as discussed in October 189: Questionnaire about Art and AI. This visual output becomes a cultural commentary on the potential and limitations of generative art to capture the richness of community, tradition, and identity, resonating with the overarching themes of AI’s role in reshaping our visual and cultural landscapes.