When the Machine Learns to Touch: KIESECKER | HOESS and the Question of Artificial Empathy

A dancer extends an arm. The gesture is precise, weighted with intention. Across the stage, something responds. Not another human body, but a presence named BUGGY, an AI assistant rendered visible through extended reality. The question hanging in the air of ZKM's Media Theater is not whether the technology works. The question is whether what happens between them can be called touch.

This is the premise of #touch, the world premiere by Karlsruhe-based choreographers Sarah Kiesecker and Dominik Höß, running November 12-15 at the ZKM | Center for Art and Media as part of TANZ Karlsruhe's 30th anniversary edition. The work sits at an intersection that should interest anyone thinking seriously about human-AI interaction: not the interface as utility, but the interface as site of intimacy.

The Venue as Statement

The choice of ZKM is not incidental. The Center for Art and Media has spent decades positioning itself at the frontier where technology meets cultural production. Its Media Theater has hosted experiments in interactive performance, generative sound, and immersive installation. When ZKM's programming lists #touch alongside the UN_CTRL Festival for Convivial Technologies and Arts, the institutional framing becomes legible: these are works that treat technology as cultural material, not neutral tool.

For KIESECKER | HOESS, the venue amplifies their central provocation. The duo, working together since 2019, has built a practice around what they describe as "democratic" choreographic processes, where all production decisions are shared equally between the two artists. Their movement vocabulary emphasizes precision, synchronicity, and physical risk. Partnering and acrobatic elements feature prominently. The body, in their work, is always a site of negotiation.

Now they are negotiating with something that has no body at all.

What BUGGY Represents

The AI assistant at the center of #touch is named BUGGY. The name itself carries a certain honesty: bugs are errors, glitches, the places where systems fail to perform as intended. To name an AI collaborator BUGGY is to acknowledge from the start that this partnership will not be seamless.

The work is described as an "intermedial XR experience," meaning it layers extended reality technologies over live performance. The dancers share the stage with a presence that exists only through technological mediation. BUGGY responds, adapts, perhaps even initiates. The choreography becomes a test case for a question that extends far beyond the theater: can artificial systems participate in the kind of reciprocal attention that defines genuine collaboration?

This is not a rhetorical question. It has implications for anyone designing systems where humans and AI are expected to work together. The language of "collaboration" saturates AI discourse, but the word often masks a fundamental asymmetry. The human adapts to the system's constraints. The system optimizes for predefined metrics. What gets lost is the quality of mutual responsiveness that characterizes actual partnership.

#touch makes this tension visible. When a dancer reaches toward BUGGY, what determines whether the gesture is received? When BUGGY responds, what makes that response feel like recognition rather than mere reaction? The difference matters. It is the difference between a tool and a presence.

Physical Empathy as Design Problem

The festival program describes the work as questioning whether "true artistic collaboration and physical empathy are possible" between live dancers and an AI. Physical empathy is a precise phrase. It points to something beyond cognitive understanding, toward the kind of bodily attunement that allows two dancers to move as one, to anticipate weight shifts, to catch each other.

This is the hardest thing to simulate. Current AI systems excel at pattern recognition, at generating plausible outputs based on training data. What they cannot do is feel the weight of another body, the micro-adjustments that happen below conscious awareness, the way breath synchronizes between partners in physical contact.

KIESECKER | HOESS are not naive about this limitation. Their practice has always centered on the physical demands of partnering, the trust required for acrobatic elements, the precision that comes only from bodies learning each other over time. To introduce an AI into this dynamic is to ask what remains when one partner cannot be touched.

The answer, whatever it turns out to be, will be instructive. Not because it will prove AI can or cannot achieve empathy, but because the attempt will reveal what empathy actually requires. Sometimes the clearest way to understand a human capacity is to watch a machine fail to replicate it.

The Festival Context

#touch premieres within a festival that has spent three decades cultivating exactly this kind of experimental work. TANZ Karlsruhe 2026 spans eight venues across the city, from the Badisches Staatstheater to the Tollhaus to the Scenario Halle. The programming ranges from international heavyweights like the Batsheva Dance Company performing Ohad Naharin's Sadeh21 to regional showcases like the Lange Nacht der kurzen Stücke.

What distinguishes the festival is its commitment to production, not just presentation. Both world premieres this year, #touch and the ANA Dance Company's The Cycle, were developed specifically for this edition. The festival functions as a laboratory, a space where artists can take risks that commercial venues might not support.

This matters for understanding what #touch represents. The work is not a tech demo dressed in artistic clothing. It emerges from a choreographic practice with established concerns, presented within an institutional framework that values experimentation over spectacle. The AI is not the point. The AI is the lens through which familiar questions about collaboration, presence, and physical intelligence get refracted into new configurations.

What the Work Asks of Its Audience

Standing in front of a work like #touch requires a particular kind of attention. The temptation is to evaluate the technology: how responsive is BUGGY, how convincing the illusion, how sophisticated the underlying systems. This is the wrong frame. The technology is infrastructure. What matters is what happens in the space the technology creates.

Notice what the body does when watching a dancer reach toward something that is not there. Notice whether the gesture reads as connection or as absence. Notice whether BUGGY's responses feel like answers or echoes. These are not technical questions. They are phenomenological ones, questions about how experience is structured, about what presence requires.

The work will not resolve the question of whether AI can achieve genuine collaboration. No single artwork could. But it will make the question tangible, embodied, available for examination. This is what art does that white papers cannot: it creates conditions for felt understanding.

The Larger Stakes

For policymakers and technologists thinking about human-AI interaction, #touch offers something valuable. It offers a test case where the stakes are aesthetic rather than operational, where failure does not mean system downtime but something more subtle: the failure to connect.

Most AI governance discourse focuses on risk mitigation, on preventing harm. This is necessary but insufficient. The question of what positive human-AI relationships might look like, what genuine collaboration would require, what it would mean for an AI system to be present rather than merely responsive, these questions rarely get asked in policy contexts. They are too soft, too speculative, too hard to measure.

But they are not unimportant. As AI systems become more integrated into daily life, the quality of human-AI interaction will shape experience in ways that extend beyond efficiency metrics. Whether those interactions feel collaborative or coercive, responsive or mechanical, present or absent, these qualities will matter. They will shape trust, adoption, and ultimately the kind of society that emerges from this technological transition.

#touch will not answer these questions. But it will ask them in a form that can be witnessed, discussed, and remembered. The artifact will hold what the discourse tends to forget: that the body knows things the mind has not yet articulated.

For those tracking how European institutions are creating space for this kind of inquiry, the Human × AI Content Hub continues to document where artistic experimentation meets the practical questions of AI governance and deployment.