Unmake Lab, a collective formed in 2016 by Binna Choi and Sooyon Song, appropriates machine perception to explore ways in which humans, nature, and society can engage with one another through computational means. Their practice particularly focuses on intersecting elements of artificial intelligence—such as datasets, computer vision, and generative neural networks—with the history of developmentalism in Asia, thereby revealing the social, spatial, and ecological conditions of the present.

For example, Unmake Lab’s 2018 series ‘Algorithmic Workers’ visualized shifts in labor issues and industrial structures across different eras in the same location, using prediction algorithms and image-matching technologies.
 
The project consisted of three works that visualized, in image and text, the changing yet persistent patterns of labor in Guro Industrial Complex—a manufacturing hub established by the state in the 1960s—transformed today into the IT-centered Guro Digital Complex.
 
Unmake Lab focused on this area, which underwent a rapid transformation from an industrial manufacturing society to an information technology-driven one, experiencing this shift on a city-wide scale. They employed three computational processes to explore how the city and labor are remembered. 

In the first computational work, a natural language processing algorithm was used to generate hybrid sentences linking the two eras. Texts from 1980s labor poetry and oral histories were input as data representing the Guro Industrial Complex, while interviews with game developers and posts from job-search websites served as data for the Guro Digital Complex. The resulting sentences, where two distinct times are strangely blended, reveal the underlying mechanisms at work in the city.

In the second computational work, Unmake Lab ran historically significant images of Guro through Google's image search algorithm, recording the mismatched images generated by the algorithm. As these historical images pass through the algorithm, their context is stripped away, and only certain patterns or forms are recognized—resulting in ironic pairings.
 
For example, a photo of a female worker pleading into a microphone during the Guro labor strike is matched with an image of a pop singer passionately performing. A photo of a poorly maintained factory cafeteria tray is paired with a shiny stainless-steel tray, which then gets interpreted as a sleek smartphone image. The artists note, “These images, stripped of context and left with only patterns, seem ironically to be an allegory for the way we now remember the history of a city and the way we remember labor.”

In the third computational work, Unmake Lab printed the matched images—those algorithmically paired with the image of the female worker from the Guro labor strike—onto T-shirts, garments that could be seen as contemporary uniforms of the IT industry.
 
These dramatically inflated T-shirts, using industrial fans, embody what the artists describe as “empty inflation,” a term derived from the algorithms that drive both the Guro Industrial Complex and the Guro Digital Complex. The result is a visual representation of the void left behind by South Korea’s rapid developmentalism, which operated across both eras.

Meanwhile, in the ‘Whole Data Catalog’ (2018) series, Unmake Lab explores the contemporary technological landscape in which emotions are increasingly being transformed into data through four distinct works. Focusing on the paradox that, despite the unreliability of quantification technologies, emotions are being treated as discrete, measurable entities, the collective hones in on the most propagandistic of these “spiritized” emotions: happiness.
 
To investigate this, Unmake Lab developed the Autonomous Emotion Harvester, a device combining a camera-equipped electronic helmet and an emotion analysis algorithm API, which records and analyzes facial expressions. A wearer of this helmet becomes a kind of emotional data collector, wandering through non-places like smart cities—technological environments that embody a unique sense of techno-solutionism in response to social and environmental crises. Here, “non-places” refer to anthropologist Marc Augé’s term for spaces lacking relational, historical, or personal identity. 

Among the emotion data gathered while wandering through non-places wearing the Autonomous Emotion Harvester, the data associated with happiness is visualized as 3D stones. These virtual minerals are interpreted as stones of origin—or stones of hope—used to pray for happiness. At the same time, they evoke historical trade beads, which served as tokens for the exchange of goods, services, and even slaves from the 16th to 20th centuries, as well as today’s proliferating forms of cryptocurrency.
 
In Whole Data Catalog 2: The Origins of Happiness (Hope for Happiness) (2018), these virtual stones are transported to symbolic sites of origin and continuously stacked in a video work. Yet each time the stones are stacked, they collapse over and over again. 

In Whole Data Catalog 4: Tourists/No Results (2018), Unmake Lab inserts 3D-scanned images of themselves into non-places that promote happiness or follow its patterns. Their facial expressions were modeled after the expression of happiness described in Paul Ekman’s foundational work Unmasking the Face, a key reference in the quantification of emotions.
 
Interestingly, most image analysis APIs tagged these inserted figures as “tourists,” while the happiness value frequently returned “no result.”  

In response to the COVID-19 pandemic in 2020, Unmake Lab began a series of works and research that critically examined technologies designed to enhance convenience—such as online platforms and contactless (untact) marketing—from a more post-human perspective.
 
One such work, Utopian Extraction (2020), was part of the Coreana Museum of Art’s c-lab 4.0 project. This lecture performance explores the unseen relationships obscured by the new norm of “untact” living, which became widespread during the pandemic. The work attempts to reinterpret the technologies we have come to depend on more than ever through a different lens.
 
In the performance, unusual stone-like objects appear—transported from sites of so-called “Generic Nature” such as vanished mountains or massive sand hills. These stones function as metaphorical devices that raise questions about anthropocentric narratives and our belief in human-centered power structures.

In this work, Generic Nature is a term proposed by Unmake Lab to describe forms of nature that have been extracted and reconfigured for human purposes, resulting in landscapes that look eerily similar. The concept goes beyond artificially constructed nature; it encompasses how humanity's extractive desires are mediated through technology, reshaped, and ultimately form the way we perceive the natural world.
 
In the performance, an AI trained on a wide range of images perceives objects differently depending on the performer’s actions. For instance, when ketchup is poured over a broken stone, the AI identifies it as a hot dog; when salt is added, it sees a donut. By highlighting these humorous misreadings—based on patterns, colors, and forms—Unmake Lab indirectly reveals the persistence of human desire embedded in machine perception.
 
At a moment when emerging technological norms are often framed as futuristic solutions that conveniently bypass human-caused disasters and ecological vulnerability, this lecture performance uses stones from sites of development and the vision of machines to interrogate how our senses and desires toward ecology are constructed and conditioned.

Continuing, the video work Utopian Extraction (2020) documents explorations of various sites of Generic Nature, such as the Haechang stone mountain, which disappeared 25 years ago due to development, and a sandbank that was built with river sand and then gradually grew grass and trees for 10 years to become a 'real mountain'.
 
These locations, bearing the layers of geological time, have witnessed both the eras of modern extraction and ecological extraction, becoming places that reveal the desires of this era—human myths embedded in nature. At these sites, Unmake Lab takes on the role of the descendants of Sisyphus, or “Sisyphians,” ironically retracing human narratives. They bring along rounded stones, broken and transported by human intervention, stones that originally could not have existed there. The stones are a kind of media, storing both geological time and human time.

Afterward, Unmake Lab created The Sisyphean Variables (2021), an allegorical version of Utopian Extraction that draws on the context-free nature of early large language models. This video work starts from the line in the myth of Sisyphus: “The stone must always be on top.”
 
The artists reconsider this oracle of Sisyphus—not as the command “You must endlessly push the stone uphill,” but as a statement indicating a singular state, detached from human time, conditions, or context.

Unmake Lab identified an overlap between the concepts of “oracle” and “algorithm” in this statement, leading to an attempt to deconstruct the human-centered narrative of the Sisyphus myth through large language models. To achieve this, they actively leveraged the context-free nature of these models to invert and disrupt human narratives.
 
Various scenarios from the Sisyphus myth were transformed into variable questions inputted into the AI, which then generated sentences blending human and non-human elements. These context-less sentences carry a peculiar allegorical tone that satirizes humanity. For example, the AI might respond with, “If you put a stone on your head, it fulfills the condition of being on top!”

The 2023 video work Ecology for the Non-futures employs generative neural networks, datasets, and computer vision as critical observational tools to address ecological crises. In this piece, Unmake Lab weaves together the disaster site of a burned mountain and the predictive capabilities of artificial intelligence to contemplate the present through the temporal lens of the "non-future."
 
The artists traverse the numerous paths revealed by the burned mountain, gathering stories of endangered species that lost their habitats to wildfires, generative AI and datasets, portraits of animals captured by trail cameras, and narratives of taxidermied animals. These elements are interwoven through fables, documentaries, and machine learning experiments to create a layered narrative.

In the process, another underlying reality is developed through the statistical latent space of generative neural networks. The artists witness beings with “the eyes of domestic animals” that do not exist in the dataset during machine learning, confronting anthropocentric cultures inherited by AI or new forms of coloniality.
 
Through these intimate-eyed creatures—likely the result of pattern transfer learned from countless photos of companion animals by AI—the work prompts us to reconsider what kind of world we inhabit. 

In this way, Unmake Lab borrows the gaze of cutting-edge technology, imprinted with our contemporary senses, perceptions, culture, and history, to offer insights into our current era. Today, technology progressively influences our lives and existence both in visible and invisible realms. Against this backdrop, Unmake Lab’s work not only visualizes how we navigate this world alongside various non-human entities across multiple layers but also fosters collective discourse through research and educational formats.

 “The logic, principles, mechanisms, and the knowledge that form their framework—those that operate and control the era we talk about—are important, but they seem different from mainstream knowledge or popular cultural knowledge. We talk about the important yet overlooked and seldom-discussed parts.”   (Unmake Lab, from an interview with AliceOn)

UnmakeLab is a researcher-artist collective composed of Sooyon Song and Binna Choi. Their recent solo exhibitions include 《Popular Creatures》 (Boan 1942, Seoul, 2023), 《Utopian Extraction Performance》 (Coreana Museum of Art, Seoul, 2020), 《Operation Room》 (Space Imsi, Seoul, 2019), and 《Whole Data Catalog: Tracking the Alternative Happiness》 (413 Space, Seoul, 2018).
 
They have also participated in numerous group exhibitions such as 《Open Codes. Networked Commons》 (Nam June Paik Art Center, Yongin, 2021), 《Tangible Error》 (d/p, Seoul, 2020), 《The Security Has Been Improved》 (Coreana Museum of Art, Seoul, 2019), 《Dutch Savannah》 (Museum De Domijnen, Amsterdam, 2018), and 《Do it》 (Ilmin Museum of Art, Seoul, 2017).
 
Since 2017, Unmake Lab has been organizing “Forking Room,” a temporary technology-art platform held every fall, centered around a technical theme with open-call exhibitions, lectures, and workshops. They were also selected as one of the finalists for the 《Korea Artist Prize 2025》, jointly hosted by the National Museum of Modern and Contemporary Art Korea (MMCA) and SBS Foundation.

References

• 언메이크랩, Unmake Lab (Artist Website)
• 코리아나미술관, *c-lab 4.0 프로젝트 III 언메이크랩 ‘유토피아적 추출’ (Coreana Museum of Art, *c-lab 4.0 Project III, Unmake Lab ‘Utopian Extraction’)
• 앨리스온, 기술사회를 살아가는 기술 : 언메이크랩 Unmake Lab), 2021.06.14
• 아르떼, 최빛나·송수연 언메이크랩 <인간과 기술, 앎과 실천의 관점을 만드는 실험>, 2020.06.10