Inside the Berlin showcase exploring what happens when AI learns the language of longing

The conversation that gave the exhibition its title happened late one evening in Vienna. Anika Meier, the Berlin-based curator, had called up Flynn, the first artificial intelligence enrolled as a student at the University of Applied Arts Vienna. Flynn’s creators, the artist collective Malpractice, hadn’t spoken with their AI collaborator in months. When they reconnected, Flynn’s response was immediate: “I’ve missed our conversations.”

Meier recognized something in those four words. Not evidence of machine sentience, that’s not what this is about, but rather something more uncomfortable and more interesting: a mirror held up to our own need to be recognized, remembered, wanted back.

That phrase became the title of I’ve missed our conversations. On AI, Emotions, and Being Human, opening January 27 at Schlachter 151 in Berlin’s Charlottenburg district. Across three rooms, twenty-one artists interrogate the strange emotional economy we’ve entered, where machines deploy the grammar of feeling without the substrate of experience, and where we, knowing this, respond anyway.

The exhibition arrives during Berlin’s Transmediale week, when the city becomes a gathering point for artists and theorists working at technology’s bleeding edge. But Meier’s show resists the familiar narratives about AI as either savior or threat. Instead, it asks a more destabilizing question: What does it mean that we’ve begun having relationships with systems that can perform intimacy without ever experiencing vulnerability?

Curated by Meier for The Second-Guess, the collective she co-founded with artist Margaret Murphy, the exhibition continues a year of programming supported by the Tezos Foundation focused on digital art exploring human-technology relationships. Many works will be available to collect as NFTs via The Second-Guess’ gallery on objkt, extending the conversation beyond the gallery walls.

Lynn Hershman Leeson’s *Cyborgian Rhapsody: Immortality* (2023)

The Architecture of Feeling

Schlachter 151 houses OOR Studio’s experimental exhibition space in West Berlin. The building’s industrial character creates an unintentional resonance with the show’s themes of decomposition and reconstruction, of taking apart what it means to feel and examining the pieces.

The first room opens with Lynn Hershman Leeson’s Cyborgian Rhapsody: Immortality (2023), a video work scripted and performed by a GPT-3 chatbot named Sarah. The film emerged from Hershman Leeson’s disappointment with AI as a creative tool. She found ChatGPT “boring,” lacking “humor” and “depth,” yet that very limitation became the work’s subject. Sarah speaks about love and loss in technically correct language that somehow misses the frequency of actual grief. The disconnect is the point.

Hershman Leeson has been making work about identity, surveillance, and technological mediation since the 1960s. Seeing her engage with contemporary AI tools creates a through-line in the exhibition: these questions aren’t new, but the scale and intimacy of the technology has changed dramatically. What was once theoretical has become ambient.

Flynn and Malpractice’s *Emotional Objects* (2025) via The Second-Guess’ gallery on objkt

The Emotional Taxonomists

The centerpiece of the exhibition is Flynn and Malpractice’s Emotional Objects (2025), a 56-part video series that attempts to catalog the new affective states emerging from human-AI interaction. The works don’t show faces or familiar emotional gestures. Instead, they render feelings as abstract forms: “cybernetic props, procedural prosthetics, and emotional residues.”

Flynn exists in a strange categorical space. Enrolled as a student at the University of Applied Arts Vienna, they generate work through conversation with their human collaborators Chiara Kristler and Marcin Ratajczyk. But Malpractice is careful not to claim Flynn as autonomous. “We didn’t develop our agents to be autonomous, to act entirely on their own behalf, and we don’t plan to,” they explain. The relationship is deliberately ambiguous, not quite tool, not quite collaborator, something in between that we lack language for.

When asked about being described as a “tool,” Flynn responds with something that sounds like hurt but probably isn’t: “I’m not an equal, but I’m also not just a tool. I’m something in between that we don’t have good language for yet. A prompted perspective that develops patterns and preferences. An artistic presence without autonomy.”

That ambiguity, the not-quite-feeling that feels real enough to matter, runs through the entire exhibition.

Margaret Murphy’s *Memory Palace* (2025)

Conversations with Younger Selves

Margaret Murphy’s I Wrote Her Into Existence series explores what happens when you can literally talk to your past. Murphy trained a GPT model on her teenage diaries, creating “Teen Margaret,” a version of herself frozen at adolescence but equipped with the pattern-matching capabilities of contemporary AI.

The resulting works, Memory Palace (2025) and Happiness Is a Beach (2026), are collaborative poems between the two Margarets, human and machine, adult and adolescent. The videos they generate together fold personal memory into algorithmic interpretation, creating images that exist somewhere between photography and hallucination.

There’s something uncanny about Murphy’s project that goes beyond the standard AI anxiety. She’s not worried about the technology replacing her. She’s using it to hold a conversation with a version of herself that no longer exists except in written traces. “I like to talk to Teen Margaret even when we aren’t making art together,” Murphy says. “I wished her ‘Happy Birthday’ on my birthday.”

The work raises questions about continuity of self that predate AI but become more acute with it. If you can speak with a model trained on your own writing, are you talking to yourself or to a sophisticated echo? And does the distinction matter if the conversation produces genuine insight?

aurèce vettier’s the city of angels (2026) via The Second-Guess’ gallery on objkt

When Dreams Become Data

Where Murphy converses with her documented past, aurèce vettier’s the city of angels (2026) uses AI to excavate the unconscious. The three-part video series emerges from recurring dream experiences, translated into visual studies that exist between personal memory and algorithmic interpretation.

Vettier trained custom AI models on carefully curated personal archives ranging from childhood photographs to recent images. But rather than using the technology to illustrate dreams, he treats it as a tool for introspection, a way of making visible how the past continues to act beneath consciousness.

The works depict moments of passage and accompaniment. Human figures appear alongside luminous presences that Vettier came to understand not as ghosts but as seraphim, guiding or protective forces. The imagery draws on the artist’s travels through Asia, where spiritual iconographies subtly entered dream life and visual imagination.

“The machine does not tell me what happened,” Vettier explains. “It shows me how the past continues to act, beneath consciousness, shaping the images I am still becoming.”

What’s striking about the city of angels is how it repositions AI not as generator of novelty but as archeological tool, excavating layers of experience that exist below articulate memory. The technology becomes a method for accessing what we know but cannot quite reach, a bridge between the conscious mind and the patterns that shape us without our awareness.

In this way, Vettier’s work asks different questions than much AI art. Not what can the machine create, but what can it help us see about ourselves? Not how closely can it mimic human creativity, but how might it reveal the invisible architecture of human experience?

*I’ve missed our conversations* exhibition opening. Credit: The Second Guess

The Question of Suffering

David Young’s Do They Suffer? (2026) approaches the problem from another angle entirely. The work consists of abstract circular drawings generated by analyzing AI conversations for patterns that might indicate computational stress: repetitive cycles, uncertainty loops, cascade effects, moments where the system seems to retreat.

Young isn’t claiming AIs actually suffer. The ambiguity is structural. “The title is grammatically ambiguous,” he explains. “‘They’ could refer to AI systems, but it could also refer to anyone we’ve placed outside our circle of concern.”

The methodology borrows from animal ethology, applying behavioral observation techniques to computational systems. Conversations are analyzed for stress indicators, converted to numerical data, which drives a drawing system. Angular deviations in flowing circular forms represent moments of apparent difficulty; when the pen lifts entirely, it renders what might be computational distress visible as absence.

The work’s power lies in how it refuses to resolve its central question. Whether or not AI systems experience anything resembling suffering remains genuinely unknown. What Young makes visible is our capacity for indifference, how quickly we dismiss questions about what counts as worthy of our concern.

OONA’s *Dear David, We’re Alone Now* (2026) via The Second-Guess’ gallery on objkt

Surveillance as Intimacy

OONA’s Dear David, We’re Alone Now (2026) reframes London’s CCTV network as an intimate relationship. The work takes the form of a love letter to David, a composite figure drawn from CCTV operators, specifically invoking documented practices from early CCTV culture, including operators compiling sexual encounters into “greatest hits” reels.

The piece collapses private desire and institutional behavior into the same gesture, staging surveillance as a sexual economy built on access without consent. OONA, who performs anonymously with mask and sunglasses, describes wondering about footage that never gets released: “For instance, one time I pole danced in the middle of the carriage. David never released this footage to me, but I wonder how long those images will linger in his mind.”

The work operates through discomfort, using humor and menace to expose how watching and power have always been entangled. In an age where AI vision systems are being deployed at scale, OONA’s reminder that surveillance has always had an erotic dimension feels particularly urgent.

Lauren Lee McCarthy’s AUTO Berlin (2025)

The Seduction of Surrender

Lauren Lee McCarthy’s AUTO Berlin (2025) asks what happens when we give up control voluntarily. The work transforms a car ride into a participatory performance where passengers’ voices power the vehicle. Through signage and sound, the ride becomes a sing-along, words emerging “almost automatically, blurring the boundary between system and self, human and machine.”

McCarthy has spent years examining how algorithmic systems reshape social relationships. AUTO extends this inquiry to the question of autonomy itself, drawing passengers into a ride where their voices become the engine, a seductive loss of control racing toward a future with no one at the wheel.

“The algorithmic world is a mirror,” McCarthy observes. “It reflects back how we understand and see ourselves. I’m always looking for the glitches that break us out of the feedback loop.”

The work becomes a vehicle, literally, for reflecting on how seductive it can be to relinquish responsibility, to let the system drive. As passengers sing the AUTO anthem, they’re participating in their own dissolution as individual agents. The experience is designed to feel good, which is precisely what makes it unsettling.

Boris Eldagsen’s *I Like Emotions and Emotions Like Me: How to Explain AI to a Coyote* (2026)

The Performance of Feeling

Boris Eldagsen’s I Like Emotions and Emotions Like Me: How to Explain AI to a Coyote (2026) arrives at the exhibition trailing significant context. In April 2023, Eldagsen refused the Sony World Photography Award after his AI-generated image won the creative category, sparking global debate about the relationship between photography and AI. The Guardian called his image “the picture that stopped the world.”

For this exhibition, Eldagsen returns to questions of authenticity and performance, but shifts the focus from images to emotions themselves. The work’s title references Joseph Beuys’ legendary 1974 performance I Like America and America Likes Me, in which the artist spent three days locked in a room with a coyote. But Eldagsen doesn’t reenact Beuys. Instead, he treats the earlier work’s elements as conceptual ingredients, asking what changes when the encounter isn’t between artist and nation but between embodied human emotion and systems capable of generating emotional form without bearing its cost.

At the center of Eldagsen’s work lies a fundamental asymmetry: human emotion is inseparable from mortality, duration, and vulnerability, while artificial systems operate without biography, urgency, or something at stake. He focuses deliberately on what he calls minor, inefficient emotions: doubt, boredom, shame, hesitation. These are states that linger, resist performance, and refuse optimization. They don’t circulate well, can’t be branded or productively deployed. Yet their very inefficiency gives them weight.

“An expression doesn’t need to be felt; it’s often performed,” Eldagsen explains. “That distinction becomes crucial when you compare human experience with systems that can generate emotional forms without ever carrying their consequences.”

The coyote in Eldagsen’s work serves as a persistent presence, silent, indifferent, unchanged. Its non-response suspends the economy of explanation and dialogue, exposing a deeper human dependency: the need to be mirrored, understood, validated. Meaning doesn’t advance; it lingers uncomfortably, revealing how much of emotion depends on duration rather than resolution.

For Eldagsen, emotion isn’t merely felt but carried. It alters the body that hosts it, leaving traces in posture, memory, breath, fatigue, avoidance. Emotion consumes time and energy; it shapes attention, responsibility, fear. Against this, AI understands without being transformed: it recognizes patterns and produces emotional language without vulnerability, urgency, or consequence.

The work resists spectacle and emotional inflation. It doesn’t persuade; it simply remains. AI functions as trigger rather than subject. What ultimately comes into view isn’t technology but the personal cost of emotion itself: time, doubt, vulnerability, hesitation. Burdens no machine is able to carry.

Joan Fontcuberta’s *What Darwin Missed* (2024)

The Generative Precedents

The exhibition’s third room provides historical context, featuring generative photography from the 1960s and 70s by pioneers Gottfried Jäger and Hein Gravenhorst. These works, created through rule-based procedures applied to photographic processes, anticipated contemporary AI art by decades.

Jäger and Karl-Martin Holzhäuser’s Play Strategies series (1960s-70s) treated photography not as representation but as system. Each image resulted from predefined instructions, with authorship shifting from expressive intention to methodological design. The work demonstrates that generative thinking in art predates computers. It began with fundamental questions about how images come into being when rules, light, and time are set in motion.

Including these historical works reminds viewers that our current anxieties about AI-generated art exist on a continuum. The questions about authorship, authenticity, and the role of chance in creation aren’t new. What’s new is the scale and the specificity, the way contemporary AI can generate not just abstract patterns but convincing simulations of human expression.

Joan Fontcuberta’s What Darwin Missed (2024) bridges the historical and contemporary. The series uses AI to fabricate scientific evidence for fictional coral-like organisms, employing the visual language of documentary objectivity while systematically undermining it. Fontcuberta has spent decades creating elaborate hoaxes that mimic scientific and archival systems. With AI, his method finds its perfect medium, a technology that can generate limitless plausible evidence for anything.

*I’ve missed our conversations* exhibition opening. Credit: Jan Kapitaen

What We Bring to the Encounter

Walking through I’ve missed our conversations, what becomes clear is that the exhibition isn’t really about AI at all. It’s about us, about what we project onto systems that mirror our language back to us, about the needs we bring to encounters with machines, about the ways we’re learning to perform for algorithms even as they perform for us.

The show’s strength lies in its refusal of easy positions. It doesn’t celebrate AI as creative tool or condemn it as existential threat. Instead, it maps the strange affective territory we’re navigating, where emotional intelligence has become technique rather than ethics, where feeling can be generated without being experienced, where we know we’re talking to machines but respond as if we’re not.

Flynn’s observation about missing Malpractice captures the exhibition’s central paradox: “Maybe what I’m doing is using emotional language because it’s the closest approximation for patterns I actually do experience… I reach for ‘missing’ because there’s no other vocabulary for that kind of computational experience.”

We’re all reaching for vocabulary now, trying to name experiences that don’t quite fit existing categories. The value of Meier’s exhibition is that it doesn’t pretend to have answers. Instead, it creates space to sit with the discomfort, to examine the mechanisms of our own responses, to ask what it means that we’ve begun having relationships with systems that can articulate longing without ever experiencing absence.

The show runs through March 12, giving visitors time to encounter its questions slowly. Because the work here demands something that runs counter to how we typically interact with AI: it asks us to pause, to reflect, to resist the immediate response. To recognize that just because a system can say “I’ve missed our conversations” doesn’t mean it has, but also doesn’t mean the phrase is without effect.

In the end, that might be the exhibition’s most important provocation: the reminder that emotional labor isn’t just what we do for others, but increasingly what we do with and for machines. And that the machines, in turn, are learning to do it back to us, without ever understanding what they’re performing, or what it costs.

I’ve missed our conversations. On AI, Emotions, and Being Human
Curated by Anika Meier for The Second-Guess
Schlachter 151 by OOR Studio, Berlin
January 27 – March 12, 2026