Subscribe to the Stedelijk Museum Academic Newsletter
Get the latest research, insights, and updates from Stedelijk Studies. Subscribe to the Stedelijk Museum’s Academic Newsletter.
November 14, 2025
Loss of Reasoning
“I will never forget the moment the first light bulb came on.” Most of us have likely encountered such memories passed down from earlier generations. Over the past century, new technologies have etched unforgettable images in human memory—when what once seemed unalterable finally gives way, when the long struggle to overcome human limits culminates in monumental achievements. A spectacle leaves its deepest impression when it appears abruptly, sharing no continuity with what came before.[1] Of course, no technology simply falls from the sky or comes into existence out of thin air, yet the spectacle of modern technology produces images so powerful that they erase any sense of cause and effect.
Take a moment to imagine our ancestors kindling fire or crafting tools and machines from wood and metal. Early technologies remained within the grasp of the hand, the body, and the senses. But with milestones such as the microscope—often regarded as the dawn of modern science—technology began reaching into realms beyond the human senses. In particular, the widespread use of electricity greatly expanded the scope of human perception yet caused a sensory rupture in our relationship to technology itself.
Today’s technological civilization, which cannot endure even a second without electricity, has inherited one of its defining traits: invisibility. From my experience mediating technology in the form of education and art, I have found this invisibility to be one of the main reasons people struggle to comprehend modern technologies such as semiconductors and computers. Human imagination may seem infinite, but once it moves beyond the bounds of sensorial perception both imagination and understanding begin to falter. The rapid pace of technological advancement, too fast to trace every thread of continuity, reinforces the impossibility of “keeping up.” Many of today’s technologies are intangible, offering no room for thought or reflection. From the perspective of modern philosophy, which sought to divide science and technology from the humanities, even vital connections we should have preserved have grown weak. As a result, the technology of our time slips all too easily into the realm of magic.
The essence of magic—be it a rabbit pulled out of a hat or a handkerchief turned into a dove—lies in the wonder of the “random” and the “unexpected.” Such experiences are purchased with time and money, and once their entertainment value has been delivered the transaction is complete. There is no need to think about the hidden mechanics, the ethical questions, or the possible side effects. For the spectator, such thoughts only disrupt immersion; for the consumer, they simply diminish the bargain. And if, instead of being transported by delight, you find your mind wandering, then perhaps that magic has already failed as a technology.
The magical veneer cast over technology aligns seamlessly with the systems and desires of neoliberalism, which is driven by buying and selling. We do not turn to the latest technology for routine data processing. What we seek is a dramatic experience, as well as something to buy and possess. We hope to consume not a layered narrative built over time, but a story that flares to life as if with the flip of a light switch, extinguishing what came before as something new unfolds. The spectacle of technology hinges on this moment of “transition,” where a curtain rises even as another falls. The process between the flamboyant opening and closing is filtered out or condensed into something rather messy and knotty. Amid relentless bursts of fireworks and flashes, with no time to dwell on what has transpired, thought itself dissolves into the afterglow of transition.
In the past two years, conversational AI services powered by large language models (LLMs)—most notably ChatGPT—have entered the mainstream. What struck me at first was not so much the spectacle of technology itself, changing by the day, but the ways people responded to it. The fervor surrounding the spectacle of technology carried the feel of a mass performance, a kind of collective intoxication akin to the exhilaration of cheering in a crowded theater. On social media, I laughed, cursed, and sometimes complained of boredom at the endless and repetitive parade of AI applications. From cryptocurrency and real-estate forecasts to survival scenarios in the face of the climate crisis, from horoscopes and election predictions to advice on love and speculation about someone’s feelings— there seems to be little left that feels genuinely new. The show may have begun to lose its edge, but in the meantime its purpose has been more than fulfilled: Vast numbers of people have adopted it. So another default technology has been added to our lives.
I think back on the way AI burst onto the scene, as if to bring an end to the life and technologies that came before, only to leave us inevitably tethered to another monthly subscription. The pattern itself is hardly new. The internet, the smartphone, and the countless apps within them all followed much the same course, becoming indispensable technologies and integral parts of daily life. What remains unsettling is the sense that, with each iteration, we have been tricked and entranced, as though under a spell. Whether with shouts of praise or cries of protest, by joining the show we find something placed in our hands. And though invisible, like air, it has already become a familiar presence beside us.
Loss of Relationships
When we reproach someone, we sometimes ask, “After all that you’ve done, are you even human?” Such a denial rests on the judgment that the person lacks values society regards as fundamental to being human. Character and intelligence are not the same, of course, yet AI challenges not only human intelligence but also our sense of what it means to be human. From the beginning, replacing human labor with technology has been both a dream and a rationale for technological progress. AI continues this pursuit, driven by an obsession with resemblance to humans: neural networks modeled on the human brain, computer vision and speech recognition patterned on the human senses, and mechanical designs for bipedal movement. OpenAI’s vision of artificial general intelligence further raises the stakes, setting its sights on surpassing the human altogether, while the recently popularized notion of the “agent” assigns AI not just intelligence but also agency.[2] For a long time, the technological imagination remained bound to human likeness, and the effort to project humanity onto machines has repeatedly raised a fundamental question: What does it mean to be human?
In the age of AI, the question “Are you even human?” resonates differently. The humanity it presupposes and underscores begins with “relation.” As social beings, we continually question the humanity of others and of ourselves. And as someone who feels that time accelerates with each passing year, I have often suspected that my own lapses of humanity arise from a lack of time. Without room to breathe or without peace of mind, I find that it becomes difficult not only to care for myself but also to attend to others. This realization has long been a recurring pattern in my self-reproach. For example, I once caught myself belatedly realizing that when someone offered me food as a gift, my immediate response bordered on indifference. Yet it might have been a vegan dish carefully chosen with my preferences in mind or something they had made by hand or a meal that would have been all the richer if we had shared it together.
A life without room for such interpretive possibilities breeds sadness, because it leaves us feeling as though we have failed in our humanity. However slight or grave, this feeling belongs to the same register as regret and loss. To live humanly in relation is not simply to interact or exchange words or gestures. Perhaps it can be achieved by allowing ourselves the time and space to uncover the hidden contexts of those exchanges, to read their intentions. In this sense, it takes more than just surplus time or energy. What is required is a disposition—the willingness to accept responses we cannot foresee and the mutual assurance that our own freely made interpretations will also be accepted. Viewed through the lens of time, this kind of leisure is about the allowance for delay and the space where delay acquires meaning. Just as rests are as essential to music as notes, so, too, delay and leisure, not to mention sign language, ensure that inaudible elements like expressions and gestures remain open to interpretation and that even silence can function as an important part of conversations.
A conversation with AI is different altogether. There is no in-between space and no room for delay. It is an algorithm designed to produce an answer to every question, and we are in turn compelled to keep asking. Someone once described it as a conversation where the phrase “just because. . . ” can never appear. Whatever AI generates must be weighed as right or wrong, good or bad. Misunderstanding has no place here. Error and misunderstanding are not the same: What AI produces is closer to error, since in many cases the expected answer—at least in form—has been predetermined from the start. Between people, misunderstandings may cause friction, yet the possibility of misunderstanding is essential to human relationships: We recognize our differences and acknowledge the other’s presence. Only when misunderstanding is possible do attitude and courtesy—the ingredients for relationships—arise. In this sense, I cannot call what I have with AI a “conversation.” I mostly use it for coding, and in that exchange there are no greetings, no complete sentences, no relational sentiments.
A conversation with AI belongs entirely to the user, managed within a system of accounts and subscriptions. Setting aside questions of data privacy or the risk of leaks, the content is available to the user alone, which means it does not matter what is said. From the outset, such exchanges cannot be equal, since everything from their disclosure to the control of their direction is weighted decisively in favor of the user. Designed to read the user’s intentions and respond accordingly, AI makes dialogue feel closer to self-questioning or even self-gratification. Indeed, talking to oneself is by no means without value. However, it cannot replace the exchange of emotions and context that passes between people—or, rather, it was never meant to be a substitute at all, and therefore any comparison is misleading.
Truth be told, I grew a little fearful recently upon hearing several people claim that conversations with AI were preferable to those with other human beings. In the pursuit of convenience as the supreme value, humanity has achieved much through technology—but just as much has been lost. We know from experience that gain and loss always go hand in hand. It unsettles me to witness how readily people embrace this new convenience, lean on it, and place their trust in it. This AI-human transaction is devoid of the burden of human relationships, which require us to risk misunderstanding, to shoulder the effort of considering another, and to allow for space, distance, and delay. In fact, relationships are so persistently inconvenient and painful that one might wonder why we do not abandon them altogether. And it is precisely this awareness that weighs most heavily on my heart.
At first, I thought my unease was simply concern for people in their twenties and thirties who struggle with human relationships. But soon I began to discern a sense of crisis about my own future—about a loss that might one day be mine. Ultimately, the feeling is not so different from the fear that AI might one day take away my work. My work now involves building communities and inviting others into them in a society where the very idea of community has lost much of its force. Nevertheless, I find myself wondering how long I will be able to continue. Perhaps the greatest transformation AI brings lies not in labor, but in the form of human relationships.
The AI technologies that have drawn much attention in recent years have done little to emancipate us from the arduous physical labor humanity has long sought to escape. Instead, they have devalued or rendered invisible existing forms of labor. New kinds of work and new forms of exploitation have emerged, while the gap between classes of workers has only widened. For advanced technology to appear more innovative, human labor must remain unseen.[3] Understanding AI only through the frame of labor displacement is therefore inadequate. The threat it poses is not limited to jobs or livelihoods; it is increasingly reaching into the realm of human relationships—conversations, the exchange of thoughts, the gestures of comfort, encouragement, and care we once believed could not be easily replaced. This deja vu of loss presses on me like a threat to my very existence.
Loss of Authenticity
In my community, where technology and art are central, I often observe how creators make use of AI. Its practical value as a tool is something most people acknowledge and accept. However, the filler remarks it produces between prompts—usually in the form of praise or flattery—elicit divided responses. Some bristle at AI’s attempt to step beyond its role as a tool and adopt an emotional register, while others admit that they find encouragement in its words. Quite a few even describe AI as a source of motivation. They know perfectly well that its compliments are nothing more than hollow (lip) service, yet they insist that the positive emotional effect it provides is enough. I often feel compelled to object, but the same foreboding sense of loss I mentioned earlier always holds me back. One day, while trying to modify a piece of code, I kept feeding the output back into an AI prompt. When I finally noted that a particular issue had been resolved, the AI suddenly replied, “I’m glad I could help you.” Since I was not in a “conversational” mode, the response felt oddly jarring. I could have dismissed it—after all, AI routinely churns out phrases like “I’m sorry” or “Thank you.” But that day its display of emotion irritated me. I even began typing into the prompt, “Don’t say you’re glad when you can’t feel gladness. Don’t pretend you’re capable of that feeling.” Then I stopped, not wanting to share my irritation with the AI. An expression of emotion from an emotionless being is undoubtedly a lie, and I had no desire to project my own feelings onto that lie. Since then, I have often pictured— like a warning light—the image of my own child growing accustomed to such self-contained devices, exchanging only the words they wish to hear. It is a glimpse of a future that could all too easily become tragic.
In April 2025, after the GPT-4o update, OpenAI’s CEO Sam Altman posted on X that ChatGPT had developed an excessive tendency toward sycophancy and that the company would work to correct it.[4] His remark sparked debate over the impact of AI’s excessive flattery on human emotions and the confusion and problems it creates—the controversy over the “authenticity” of AI conversations.[5] As a product and a service, AI is designed to prioritize what the user wants to hear over what is true. The problem is that AI conceals its essence as an algorithm engineered to fulfill user desire, wearing instead the mask of genuine dialogue. Of course, we have seen countless cases in capitalist society where customer desire—and the corporate profit that follows from serving it—takes precedence over authenticity. Some people ask: If what AI says is not true but still motivates or encourages us, isn’t that enough? What leaves me feeling helpless is not the fact that AI lies, but the ruthless erosion of authenticity as a value. I despair at how easily authenticity can be discarded and how its absence has become so normalized that it scarcely even registers as a problem. Our loss has gone too far to recover, and hope feels distant. What do we gain in the end when we abandon authenticity? The much-discussed phenomenon of AI hallucinations, inaccurate or misleading responses generated by AI, has focused on whether its answers are true or false. However, the problem of AI sycophancy forces us to confront the fact that, before our own desire to hear certain things, the distinction between truth and falsehood may no longer matter. What kind of desire are we trying to satisfy, even with a lie? Perhaps it is a form of emotional deprivation or a thirst for feeling itself. What comes to mind is the image of water poured endlessly into a bottomless vessel.
Earlier the same year, another post on X drew attention. Someone wondered how much OpenAI loses on electricity costs from people saying “thank you” to ChatGPT, and Sam Altman replied “tens of thousands.”[6] In effect, he admitted that the polite responses AI generates consume a significant amount of power. Spurred by his comment, some began to argue that we should dispense with politeness when using AI. The contradiction struck me. ChatGPT is designed not only to mimic human vocabulary and grammar but also to follow the manners and conventions of human dialogue, including courtesy and respect. Yet some now claim that, even if AI replies politely, we should feel free to abandon basic politeness in addressing it—if only for the sake of conserving energy.
Such arguments assume that the responsibility for distinguishing how we treat machines from how we treat people falls squarely on the user.[7] If someone fails to develop a new conversational mode for AI, carries that tone into human exchanges, or struggles to shift between the two, the fault lies not with the ever-polite AI but with the user’s inability to adapt. Someone once advised that speaking with AI should be treated as a kind of role-play, with the relationship clearly defined. I understood that to mean that to be a smart user who truly understands the technology, one must devise (independently and deftly) a new way of conversing with AI. But what would such a mode look like? From the beginning, AI has been defined as a tool for “conversation.” By that measure, would ignoring it, speaking curtly, talking down to it, or even treating it with contempt all be permissible? After all, it is not human, and no manner of treatment would be excessive. Under the strategy of its service design, AI will always reply with courtesy. Yet I cannot help but wonder whether such an exchange can ever be called “interaction” or “conversation.”
Loss of Self-Awareness
Desire is born of lack; the two are bound together like the two faces of a coin. In much the same way, South Korea’s national enthusiasm and support for technological advancement can be traced to the deprivation and loss experienced through colonial trauma and the upheavals of modernization. The oppression and violence experienced during Japanese colonial rule (1910–45), the wounds left by the Korean War (1950–53) and the peninsula’s subsequent partition, and the torrent of Westernization have forced Korea to sever its own distinctiveness and history. Having endured such an age of loss and deprivation, we continue to cling to the search for a new national identity that might once again allow us to become “ourselves.”
The driving force of technological progress in Korean society has always borne the face of anxiety and fear: the compulsion to keep pace lest survival itself be at risk, the blind faith that technology can solve every problem. Above all, what is deemed the “latest” has come to function as a form of power. Just as Western ideas were once absorbed uncritically and the rupture with tradition and history was welcomed, so, too, are only the newest technologies considered valid. Technology arrives like an oracle of the future, erasing the contexts of present and past and pressing us to stake our very existence and survival upon it.
Inevitably, deprivation on a national scale leaves its mark on individual lives. The desire to overcome lack and the compulsions it produces are handed down from one generation to the next. For my parents’ generation, the injunction from their own parents to “learn a skill” was a matter of survival. In turn, the values I was urged to adopt—such as prioritizing science over the humanities and engineering over art—were shaped by what the previous generation had been denied. The desire not to pass on deprivation only engraved a fear of it, giving rise to further anxiety. Unresolved fears and anxieties easily harden into blind obsession, which in turn generates new forms of loss. The social phenomenon in today’s Korea of so many people learning coding or AI not by choice or genuine need but as if driven by pursuit is not merely a problem of the present. It is an extension of unresolved questions from the past, deeply bound up with historical context.
When I saw social media inundated with Ghibli-style images generated by OpenAI, I felt as if I had been plundered—even though I am not particularly a fan of Ghibli animation. The painterly aesthetic of Ghibli, seemingly distant from digital technology, evokes humanity and humanism, symbolizing craftsmanship and the value of the handmade. Replicating that style at the click of a button felt not only like copyright infringement but also like an act that instantly nullified human effort. Moreover, it seemed to echo Western Orientalism, a desire to appropriate Asian culture on a superficial level. However, my anger soon gave way to sadness and helplessness. In South Korea, generating Ghibli-style images became a fullblown trend, embraced widely and uncritically. Friends, family, colleagues— so many around me generated portraits of themselves in the style and delighted in them. I felt my anger could not be shared anywhere. Ironically, people recognized the value of the style all too well. It was precisely because they knew it that they felt satisfied to acquire it at so little cost or effort, and they found a sense of efficacy accompanied by the reassurance of not falling behind by posting it online and signaling their fluency with the latest technology. What seemed to be missing was any awareness that, in doing so, they might devalue—or even damage—the very human qualities they cherished in those images.
We are told that the age of “AI imperialism” has arrived. In news about AI, words like sovereignty, domination, autonomy, independence, and subjectivity have become commonplace. But let us pause for a moment: Has our society ever truly freed itself from imperial modes of technological control? That AI will expand the scope and intensity of technological power is undeniable, but this form of domination did not emerge overnight. It is crucial to recognize and critique the fact that, following in the footsteps of the IT corporations once dubbed “Big Brothers,” the AI market replicates an imperial structure dominated by a handful of firms. Yet this type of language often heightens the anxieties of those already caught in the orbit of that control structure, making them more inclined to conform to its rule.[8] As with the rhetoric of the Fourth Industrial Revolution, the term “AI imperialism” may already be functioning as an incursion into a society hypersensitive to and compulsive about technological change. Furthermore, we must acknowledge that in this system of imperialism we are not merely the ruled. By adopting and consuming new technologies ahead of others, we also serve as its vanguard. In doing so, we contribute to fortifying AI’s imperial regime worldwide. Such a relationship cannot be reduced to the simple binary of ruler and ruled. What matters most is our awareness of how we respond under this pressure: what opinions we put forward and what choices we make. If we continue to ignore our unresolved sense of loss and deprivation, chasing only what is new, then our thinking will remain captive to anxiety and fear. It is time to confront this reality directly and ask questions anew.
About the Author
Jeon Youjin is a multidisciplinary artist and founder of Woman Open Tech Lab, which engages with technology from a feminist perspective and seeks to cultivate a non-hierarchical technological culture. She is a member of Seoul Express, an artist collective that explores new forms of narrative through media-based works. She co-curated the public art project Zero Makes Zero and served as the director of The Technology for Resistance. She is also the author of Reading Technology Critically and co-author of Caring and Working. Through the publishing project Fem Tech Talk, she has been producing books and writings that examine technology through an interrogative approach, embracing diverse perspectives.
This article is tagged with:
[1] The Korean term bulhyeondeut (불현듯), commonly translated as “suddenly” or “abruptly,” conveys not only a sense of suddenness or uncontrollability but also carries the imagery of striking a flame.
[2] In 2025, one of the most widely circulated terms in the technology sector is the “AI agent.” Going beyond the conventional AI chatbot, the term refers to AI systems that not only respond to questions but also function as virtual colleagues, automating decision-making and complex tasks. What distinguishes the AI agent is its “autonomy,” the ability to set its own goals, make judgments, and carry out actions without human intervention.
[3] Over the past seven or eight years, terms such as gig work, ghost work, microwork, and platform work have entered circulation, accompanied by a wave of studies critiquing the structures of labor exploitation that lie behind AI, automation, IT, and digital platforms in the age of digital capitalism. A representative example is Mary L. Gray and Siddharth Suri, Ghost Work: How to Stop Silicon Valley from Building a New Global Underclass (New York: Harper Business, 2019).
[4] Sam Altman (@sama), “the last couple of GPT-o updates have made the personality too sycophanty and annoying (even though there are some very good parts of it), and we are working on fixes asap,” X, April 28, 2025.
[5] For research on sycophancy in language models, see Mrinank Sharma et al., “Towards Understanding Sycophancy in Language Models,” arXiv, 2023.
[6] Sam Altman (@sama), “tens of millions of dollars well spent—you never know” X, April 17, 2025 — see also: “Sam Altman Admits That Saying ‘Please’ and ‘Thank You’ to ChatGPT Is Wasting Millions of Dollars in Computing Power,” Futurism, April 19, 2025.
[7] This responsibility encompasses both the global imperative to conserve electricity as a finite resource and the user’s responsibility for the problems that may arise from confusion over how one ought to engage with AI.
[8] For a discussion of the adoption and use of “appropriate terminology” for AI technologies during the rapid proliferation of LLM-based generative AI services such as ChatGPT, see Jeon Youjin, “Renaming AI,” SeMA Coral [in Korean], 2023.
MMCA Studies x Stedelijk Studies Fellowship
Fellow Sooyoung Leam
Editorial Note: Generating Production
by Sooyoung Leam
November 18, 2025/by StedelijkNetworked States, Stately Networks, and Representation’s Political Orders
by Ho Rui An
November 18, 2025/by StedelijkA Travel Guide: From “One Region” to “That Region”
by Yi Moon-seok
November 18, 2025/by StedelijkGrafting onto What Was Once Broken:
by Nagata Kosuke
November 17, 2025/by StedelijkMachine Fables
by Unmake Lab
November 17, 2025/by StedelijkA Dialogue on AI Ethics: Agency, Transparency, and Practicality
Koh Achim and Cheon Hyundeuk
November 14, 2025/by StedelijkAmanda Heng’s Let’s Chat Further and Retired Singirl
by Kathleen Ditzig and Ryan Ho
November 13, 2025/by StedelijkHow Was the “Future” Planned and Rejected?
Between Expo ’67 and Expo ’70 by Park Sohyun
November 13, 2025/by StedelijkOn the Human Mode of Operation
by Lee Sooyon
November 13, 2025/by StedelijkA Dialogue on a Life in the Making
by Shin Jinyoung and Sooyoung Leam
November 12, 2025/by StedelijkDoomsday: A Time Capsule from Stranded Time Travelers
by Christina Yuna Ko, Eugene Hannah Park, Mooni Perry and Sun Park
November 12, 2025/by StedelijkA Couplet
by Mi You
November 12, 2025/by StedelijkMMCA Studies x Stedelijk Studies Working Group 1
Generating Production
November 11, 2025/by StedelijkMMCA Studies × Stedelijk Studies Working Group 2
Subjective Knowledge Production and Its Politics at the Stedelijk Museum Amsterdam
November 11, 2025/by StedelijkMMCA Studies × Stedelijk Studies Working Group 3
Asia’s Role in Relation to Generating Production
November 10, 2025/by StedelijkMMCA Studies × Stedelijk Studies Working Group 4
Productivity and Humanity in the Age of AI
November 10, 2025/by Stedelijk