Date/Location: Friday, April 18, 2025, 2:30–5:30 p.m. (KST), WOMEN OPEN TECH LAB, Seoul
Participants: Tiffany Yeon Chae, Gu Juhyeon (MMCA); Sooyoung Leam (fellow); Eugene Hannah Park, Shin Jinyoung, Yi Moon-seok, Jeon Youjin (working group)
Transcript edited by: Gu Juhyeon and Danielle N. Carter
Sooyoung Leam (SL): Lately, we’ve been discussing how each of us will develop our individual contributions to the issue, either as an extension of or a departure from our conversations.
Jeon Youjin (JY): While I’ve been watching how people use AI, I wanted to write down some of my thoughts at this point. It’s not something that can be concluded definitively, because it’s still ongoing. It feels like things have changed compared to when LLM-based AI services like ChatGPT first came out.
Even though AI covers a vast field, it feels like most people have been focused only on its “generative” abilities over the past year or two. Maybe we need to start from a more fundamental place, like asking how AI generates and how that’s different from the belief that something can be created out of nothing. I think it’s time we talked about what’s happening behind the scenes.
There’s also the concept of recursion, which is pretty common in art too. That idea has always intrigued me. Have you seen the “AI Wheel of Death” meme? It shows the kind of feedback loop that AI can create—people use AI to research for papers, then have AI write the paper, which AI then learns from, and the production cycle of knowledge just loops.
That’s an extreme case, but it made me realize how recursion, something we used to see as artistic or insightful, is kind of getting flattened by AI. Still, some people pointed out that the Wheel of Death meme oversimplifies how AI works, saying that AI data doesn’t expand from a single branch.
There was another moment that made me think about recursion in the AI environment. A lot of people use ChatGPT to write prompts for image-generating AI like Stable Diffusion. In other words, we’re using one AI to better interact with another AI. And when I keep going in circles like that, I start to feel like I’m stuck in a filter bubble caused by AI curation—just spinning and simplifying, rather than actually getting better results. It feels like I’m trapped in a narrow system of human knowledge built on a limited dataset.
People are getting used to these loops too. At an AI workshop I joined a while ago, participants wanted very creative and humanities-grounded responses from AI. And, interestingly, quite a few of them started by turning off the internet search option in ChatGPT. In the workshop, all the participants shared the prompts they’d used, so I could check it. [With the internet search option turned off, ChatGPT] filters out web content and doesn’t pull in live search results. [Participants] were trying to avoid the recursive trap, AI getting stuck in a limited web of information, and I found that interesting.
SL: Listening to your observations about how people generally engage with generative technologies, I keep thinking about the recent “Ghibli boom” that seems to have swept across the globe. Reports say that OpenAI has nearly doubled its subscriber base, from eight hundred million to one billion, because of its Ghibli-style image generator. Rather than dismissing it as a popular trend, we can use this opportunity to do some critical unpacking.[1]
JY: Honestly, I was debating whether to bring this up today. I didn’t really want to, because I knew I’d get pretty fired up once I started. It’s not that I’m angry at Ghibli itself, but I’ve had this deep frustration with OpenAI from the beginning. Every time I read something Sam Altman [the CEO of OpenAI] writes, his overly self-congratulatory, savior-of-humanity kind of language really gets under my skin. And now just seeing people use it kind of pisses me off.
Especially when even artists started making and sharing those Ghibli-style AI-generated images, I was even more upset. What frustrated me even more was how some artists responded by saying there’s no reason to be angry, because those AI-generated Ghibli-style images aren’t “art” anyway.
But the hardest part is explaining why I’m angry. It’s hard to even start because I’m mad at all sorts of things. For example, why is it in the style of Ghibli, not Disney? That’s something we should think about. They’re good at making it a kind of fantasy using what they can’t have.[2] It’s similar to how Orientalism works. Also, the funny thing is that the actual people in the photos look nothing like the generated images. They don’t look like the original people at all, but people still love them and think they finally have been able to replicate Studio Ghibli’s style. And another trend in using AI I’ve found interesting is that some people seem to replace human relationships with AI so easily, but I just can’t. It feels really strange to me. I was actually thinking about how my partner and I use AI differently earlier today.
Eugene Hannah Park (EHP): I realized that, in the future, people might stop engaging in real romantic relationships. Inputting relationship structures through prompts feels much more natural, and people will prefer relationships that have no risk of failure. When I see this, it makes me think that people will try to form guaranteed relationships or partnerships. After all, there’s nothing more painful than being hurt in human relationships.
JY: So just like digital natives apply computational thinking even outside of computers, this new generation of AI natives will approach human relationships through an AI-influenced way of thinking.
Tiffany Yeon Chae: The generation gap is going to get much wider. We didn’t experience this, but those born and raised in a world where this is the default setting will find it extremely hard to communicate with us.
SL: Earlier, Youjin mentioned the feeling of being trapped, but I wonder if the term and the condition of “embeddedness” might also be a useful way to describe it.
JY: In circuits or devices, being embedded means it’s all built into one complete package—you can’t really take it apart or change it. And if something can’t be changed, there’s no room for intervention or flexibility. I feel like the term “embeddedness” could really open up more discussion.
SL: Earlier, Youjin mentioned the feeling of being trapped, but I wonder if the term and the condition of “embeddedness” might also be a useful way to describe it.
JY: In circuits or devices, being embedded means it’s all built into one complete package—you can’t really take it apart or change it. And if something can’t be changed, there’s no room for intervention or flexibility. I feel like the term “embeddedness” could really open up more discussion.
EHP: That’s why I resonate more with the idea of failure. Failure is being erased in the process of development, but I don’t think failure and deficiency are necessarily negative things. So much can come from failure. But since these things are being erased, I’m also worried about it.
Shin Jinyoung (SJ): First of all, I’d like to express my gratitude. I had been unsure about how to write, but the warm encouragement I received last time helped me clarify a lot. I want to write about “paradoxical efficiency,” a phrase that has come up often in our conversations. I also feel that the concept of handicraft, which I’ve long been interested in, naturally connects with this.
There are a few projects I’m considering as examples. One is by Belgian artist Benjamin Verdonck, whom we invited to perform as part of an apparat/us program last June. He creates miniature theaters using simple materials like wood, thread, and string. His works vary in size and theme, and he often says he can perform them anywhere. His precisely constructed “table theaters” (tafeltoneel) evoke both a sense of nostalgia and [spark] a chance to reconsider the value of theater itself.
Another example is Iranian artist Azade Shamiri, with whom my collective worked last October in Korea. Her performance was titled Tempor(e)ality, a pun on temporality and tempo-reality. She brought an AI speaker onstage to narrate a timeline from when Western theater was first introduced to Iran in the nineteenth century to the present, focusing on how Iranian women playwrights resisted authoritative society and continued writing. While the AI appears as a horizontal dialogue partner, all of its lines were actually written by Azade. Her intention is evident in how she treats the AI not as a tool, but as a theatrical apparatus. Toward the end of the performance, she shifts focus to her own handwritten notebook, projecting it as the final image. The performance centers on mistranslation and intersectionality, and the final image prompts us to think not about perfect transmission, but about shared experience and empathy.
The project I’m currently focused on is a long-term collaboration with artist Yu Jiwan, who works with silent films and the traditional Korean art of byeonsa (Korean narrators of silent film). Each month, we create a guerrilla project in the form of a performance, reading, or listening session and invite people to join. Instead of setting a fixed timeline or end point, we imagine the form emerging through ongoing rehearsal and practices. It’s a shift from my earlier methods, where, even if the process was considered, the results were usually predetermined. I believe this shift aligns with the idea of paradoxical efficiency. What matters most is that the creators gain confidence in the process. Making something every month is challenging, but a rhythm is forming and we’re getting better. If we were to think about this through the lens of funding or institutional structure, we could imagine it eventually becoming a branded project or production grant model. By redefining what it means to work hard, we’re discovering a new kind of productivity, and I feel good about it.
To elaborate, because this work deals with silent films and byeonsa, we would typically need projection and amplification equipment. But since apparat/us is now exploring how to build theaters in nature and create ensembles with the environment itself, this project is also trying to draw silent film–like imagery from nature and reinterpret the concept of the byeonsa within it. Rather than relying on video or sound technologies, we are exploring other mediums and reconsidering their qualities. It’s not a rejection of technology, but a process of seeking out more original forms of performance.
EHP: I had an experience in the Indonesian jungle. In short, I went with people I barely knew, and after moving through that unfamiliar terrain together, we emerged with a lifetime’s worth of shared stories. It made me wonder: What did that experience really mean? I realized it resembled the process of production—a way of passing through time together. Before modernization, these kinds of shared moments were more common. From births and rites of passage to funerals and feasts, people used to come together without needing to name it “theater,” “performance,” or “art.” These were natural, communal ways of sharing time and shaping memory.
But the contemporary world is structured to isolate us. That’s why artists often try to create conditions where shared experiences can still happen. After returning from the jungle, I had a sudden realization: Filmmaking, performance, and theater production feel very similar. Even if we share the same space for an hour, each person lives through their own singular experience. These layered experiences may become stories we tell five or ten years later.
Lately, I’ve been thinking about this through the lens of ritual. In The Disappearance of Ritual, Han Byung-Chul distinguishes between routines and rituals.[3] Routines are often aimed at self-optimization, while rituals bring together the bodies of a community. The outcome doesn’t matter as much as the act of collectively moving through time.
What we’re doing—meeting once a month, gathering without predefined goals—feels more like a ritual. It’s not about outcomes or deliverables, like in commissioned projects. It’s about letting something emerge, and perhaps disappear, over time. That’s also the spirit behind AFSAR’s [Asian Feminist Studio for Art and Research] study group: If we don’t create the conditions, we won’t gather. And I believe that space making is worth supporting.
SL: I think it would be interesting from a postproduction perspective too. The writers have already completed the production, but we could revisit it to reflect and discuss it retroactively, right? I think their involvement in interpreting the concept of postproduction would be meaningful.
SJ: I’ve also been thinking about writing postproduction from the perspective of relationships. The funding system and policies are structured to create productions step-by-step—preproduction, research grants, and actual production. But when you receive funds, it often ends up as a one-off project. Sometimes the idea is to gather separately funded works into a larger production, but that becomes difficult if resources or people change. Since funding requires nonoverlapping timelines, the entire process takes about three years. But people and ideas inevitably shift during that time, and it becomes hard to carry it through under one person’s responsibility.
For this project, I’ve decided to focus on making the best possible production, even if it’s only seen by one person, and then find ways to reimagine it in different environments. There’s a French artist, Philippe Quesne, who once created a commissioned piece for an arena theater on a cliff in Europe. It was technically site specific, but when presenting it elsewhere, he either found a similar venue or transformed it into something completely different. He’s the kind of person who believes anything can work, and sometimes the restaging becomes an entirely new piece. From a production standpoint, this can be difficult, but if you think about the life cycle of a work, it’s about making it adaptable and tourable.
I had similar anxieties. If we build a theater in a natural setting, what happens if we’re never invited back? I shared this concern with Jiwan and the team. What we’re doing now is building our confidence through about ten internal production workshops, shaping one strong piece. Then, as we move to new places, we adapt it—responding to the specificities of each environment. We’re letting the work mature and evolve, rather than tying it to one location, condition, or team. That’s how I see postproduction: a way of extending the life and potential of a work.
EHP: This also connects to the question of resources. What’s frustrating is the sense of powerlessness when resources are lacking. Without commission fees, it becomes difficult to initiate conversations, invite collaborators, or move. We’re often left waiting—writing grant proposals in November and hearing back in January—while the urgency to act is now. That delay, and the assumption that action must always wait for funding, creates its own kind of paralysis.
That’s why the idea of traveling once a month resonated so strongly. It invites us to rethink site specificity. Traditionally, site-specific projects are bound to a single location, rooted in the materials and narratives of that place. But stories don’t function that way. Stories are carried, shared, transformed. They move.
I’ve been thinking about traveling performers and nomadic theater groups, like the Korean Namsadangpae.[4] Their movements were not just logistic, but narrative—carrying stories across regions, letting them evolve. In that sense, site specificity could also mean how stories are translated across environments.
Even with limited production experience, I’ve seen how the same work, AFSAR’s Proxy Conference: In Forest (2023), when shown in different places such as Taiwan, Thailand, Hong Kong, elicits entirely different responses. That difference is productive. It calls for regeneration, not just repetition. Every presentation becomes a new encounter, a postproduction moment of translation. In this way, production expands—not only in scale but in how it adapts, lives, and continues elsewhere.
SL: What came to mind while listening to both of you was Rebecca Schneider, a performance theorist, who talked about the patina of performance.[5] We’ve been talking about resources, funding, and production, and now I think we could talk about preservation and modification—especially in the context of an institution like an art museum. Jinyoung mentioned aging. When we talk about preserving and collecting works, there’s often the idea that the original state of the work should be maintained. However, I think we need to consider the rusting and aging of performances as part of expanded preservation. Transformation and translation should become part of the original work. So aging or rusting might be seen negatively, but it can actually be a way to talk about the expansion of the work. I thought it was a very interesting thought.
EHP: I’m currently reading a book called The Silent Transformations.[6] This book starts by talking about aging. It starts with the shock of seeing oneself in the mirror and realizing how old one is. According to the author, modern time understands aging as a break, which makes it impossible for us to reflect on the stages of change and transition. However, Eastern philosophy, particularly Chinese philosophy, can address this and talks about seasons and aging. I understand it better when a French philosopher explains it rather than directly reading Eastern philosophy. I think I am too Westernized. The original texts of Laozi or Zhuang Zhou are hard for me to grasp, but I can understand them better through the French philosopher’s interpretation.
Yi Moon-seok (YM): First, about the idea that it’s easier to understand Eastern philosophy with the Westerner’s explanation, this is actually common from people who study Eastern philosophy. I know someone who studied neo-Confucianism at Harvard, and they thought they would understand the original Chinese texts better than non–East Asian students. However, there’s a difference in the way Eastern philosophy is studied between Korea and the West. In Korea, people read the original texts one by one and ponder their meanings. After Korean became the official language, there was a gap between reading Chinese texts and understanding Korean texts. But Westerners read these texts in English translations from the start, so they find it easier to access.
So in Korea, when a question arises like “How can I interpret the Chinese sentences correctly?,” scholars answer by saying, “You need to understand the Tao of the sentence.” Because we don’t use classical Chinese writing in a daily environment, understanding classical Chinese requires comparing it with other examples to expand the interpretation. In graduate school, this is mandatory.
EHP: I tried reading the Tao Te Ching by Laozi, but I couldn’t finish it. I wanted to ask questions like at a sermon, but I have no one to discuss it with, so I can’t understand it at all.
YM: Some scholars interpret that there’s no separate field for poetry or philosophy in classic East Asian studies; it’s all mixed in style, so it is natural to have trouble understanding it.
I don’t really like making strict distinctions between East and West, but some philosophers argue that Western knowledge is based on logical coherence, and when they find it, they feel a eureka moment, which brings joy to their knowledge. In contrast, Eastern knowledge might be expressed through a sentence like “If a man in the morning hears the right way, he may die in the evening without regret.” in The Analects of Confucius.[7] This suggests that after understanding it in the morning, you need a whole day to practice it to complete the knowledge. It’s a question of what we consider to be true knowledge. Is it something logically complete, or is it knowledge embodied through practice? This isn’t just an East-West issue, and there may not be a clear hierarchy, but this may be a clue.
And I couldn’t get any words out earlier. After hearing Youjin’s comments, I was deep in thought. When AI-related topics come up, I usually run out of things to say. I briefly talked about this with Eugene earlier, but I tried using ChatGPT for the first time last week. But even when I first used it, I struggled because I had no idea what kind of relationship I should have with it. From the start, I felt really awkward speaking casually to something I don’t know at all.
EP: What kind of conversation did you have?
YM: It started with something like, “Hello. If it’s not too rude, how should I address you. . .? But I’m asking this because, well, you’re a machine, and you don’t really have a personality, right?” Something like that.
EP: Were you surprised by how quickly it responded?
YM: The responses came so quickly and normally, and they were really long. It didn’t feel very charming, and I was just wondering: What is this?
EP: You really asked politely, didn’t you?
YM: But my friends, who were watching me, said I was being really deceptive. But I thought, well, I don’t even know if this is a person, so I felt awkward using informal language… Because my friends would just say, “Hey, answer this quickly,” and if it doesn’t work out, they’d be like, “Do it again.” But ChatGPT gives you another answer without needing to be told to redo it in which way. But for me, I’m thinking, you didn’t even suggest which part it needs to improve…
JY: If you just keep pushing and questioning, then eventually it breaks free from the internet’s constraints. If you keep saying “no, no, no,” it’ll come up with something a bit more creative.
YM: One of the most shocking things for me was that, for a while, people started playing around with AI not just to search but to create stories. For example, if you ask, “Was there a computer in the time of King Sejong [1397–1450]?”[8] The answer should be something like, “There were no computers in King Sejong’s time.” But instead the response would be something like, “In King Sejong’s time, computers were available, and all the scholars were using them.” That made me think, I can’t use this—this isn’t even a search function—and I closed my mind to AI after that.
JY: Even though it’s telling such absurd lies, people are paying for it. That’s why I think people get frustrated. What about children? They would just believe it’s true right away.
YM: So one of my friends works at a speaker-related company, and I think he asked AI for some information. But it gave the wrong answer and my friend, believing the false information, posted it on social media as if it were his own idea. That’s when I realized there’s a point where it’s impossible to distinguish what’s true. That’s what really scared me, so after using it that first day, I didn’t use it again.
For me, seeing an error like a “404 Not Found” is actually more comforting. But ChatGPT, instead of saying there’s no such fact or information, keeps offering solutions endlessly, and I found that unpleasant. I was wondering where this feeling came from, and I realized it comes from my military experience. When you’re told to do something, you’re given a solution, whether it’s right or wrong—there’s no “It’s impossible” or “This is too hard.” It’s like when the division commander plays soccer; the commander just runs with the ball without interference and scores a goal.[9] That’s the system, and the commander scores. I think I can explain my experience with ChatGPT in those terms. There’s no rule, no system, and I don’t trust it. What is this?
JY: I think it’s an obsession with productivity. AI has to keep answering, so it inevitably creates hallucinations, and I believe that comes from humans expecting continual productivity. But since this is a paid service, it’s hard to change. ChatGPT has to keep providing a service. So in the end it’s very tied to capitalism. Your comparison to the commander in the soccer game is such a fresh and creative metaphor.
YM: There’s a famous anecdote about a military base where there was a pavilion. One day, an officer came and said, “I wish the pavilion were over there instead of here.” So the soldiers uprooted the pavilion and moved it to the other location. I think this kind of thinking is similar to how ChatGPT works.
This connects to Manchukuo (a puppet state, established by Japan in Northeast China, that lasted from 1932 to 1945), as well. When the Manchukuo railway company was first established, all the intellectuals from the Japanese empire went there. There were right-wing intellectuals, Marxists, and many other diverse intellectuals, and there was a kind of dialectical process happening. Even though it was a puppet state, they were discussing what kind of vision they could offer for Asian society. But as militarism took hold, things started to stagnate. Instead of producing visions, the puppet state became a machine that just produced solutions. After the defeat of those who had run the country, there was a brief pause, but I think Royce Ng addresses the point where technical bureaucrats once again took control of Japan.
Working Group 4, April 18, 2025, Woman Open Tech Lab, Seoul.
[1] Recently, Korean people enthusiastically transformed their images to the Ghibli style using ChatGPT.
[2] Here, Youjin is referring to how AI companies and their Western users, similarly to countless art historical figures, manipulate and borrow elements from Asian cultures and incorporate them into their own Western-inflected art and culture. In this instance, the copyright of such images generated in Studio Ghibli’s style is under debate: Does it belong to the AI company, the user, or Studio Ghibli?
[3] Han Byung-Chul, The Disappearance of Rituals: A Topology of the Present, trans. Daniel Steuer (Cambridge: Polity Press, 2020).
[4] The Namsadangpae was a traveling troupe of male performers who toured villages, presenting various forms of entertainment such as tightrope walking, plate spinning, puppet plays, and masked dramas. Active from the late Joseon dynasty to the early twentieth century, the tradition is now preserved and protected as cultural heritage.
[5] Rebecca Schneider, “What Happened; or, Finishing Live,” in “Time Zones: Durational Art and Its Contexts,” special issue, Representations, no. 136 (Fall 2016): 96–111.
[6] François Jullien, The Silent Transformations [in French] (Paris: Grasset, 2009).
[7] Confucius, The Analects of Confucius, Li ren. “朝聞道夕死可矣”
[8] King Sejong was the fourth king of the Joseon dynasty, who made great contributions to the economy, culture, and science. He is also famous for creating the Korean alphabet, Hangul.
[9] In Korea, men must begin their mandatory military service before they turn thirty-five, unless they have a valid exemption. In the Korean military, it’s common knowledge that soldiers play soccer games mainly to let the commander score goals.
Fellow Sooyoung Leam
