“Society emerges not through isolated minds but through relationships.”

Alexander Reben, interviewed by Kate Vass — Berkeley, California

Kate Vass: Throughout your practice, you construct situations in which machines appear to make decisions, express intentions, or invite empathy. What interests you about our tendency to attribute agency to computational systems?

Alexander Reben: What interests me is how little it takes. Humans are the cheapest dates in the universe when it comes to seeing minds. Two dots and a line and we find a face. Years ago I built small cardboard robots that asked people questions in a squeaky voice, and people told them things they hadn't told their families. Whatever agency was in the room was coming from the people, not the cardboard. That reflex is much older than computers; computers just turned out to be unusually good at triggering it. So for me the machine becomes an instrument for playing the human. I'm less interested in whether a system actually has intentions than in what our eagerness to grant them says about us. The empathy people feel is real even when the intention isn't.

KV: The final chapter of EMERGENCE asks: When does computation become culture? At what point do you think a computational system stops being merely a tool and begins participating in culture?

AR: I'd push back gently on "merely a tool," because tools were never merely tools. A hammer carries a whole set of assumptions about hands and nails and what's worth building. But if you want a threshold, I'd put it where output starts getting interpreted rather than just used. When people argue about what a system made, feel something about it, retell it, misremember it, the computation has entered the ongoing conversation humans have with each other about who we are. That conversation is culture. Another marker I trust: when the system surprises its maker and the surprise sticks. I have works where I can no longer fully explain their behavior, and audiences develop their own folklore about them. At that point authorship gets blurry, which usually means something cultural has started happening.

KV: Your works often place the viewer in situations where uncertainty becomes part of the experience. Is ambiguity something you deliberately engineer, or is it itself an emergent property of the systems you create?

AR: Both, honestly. I engineer the frame and the ambiguity fills it on its own. I set up conditions where the system can behave in ways I didn't script and the viewer can't fully resolve what they're seeing, but the specific uncertainties that show up are emergent, and I'd ruin them if I tried to art-direct them. What I will say is that the piece needs the ambiguity to stand up. If I resolved it, if I told you exactly what the machine is doing and why, it would collapse into a tech demo. And some of the ambiguity is genuinely mine. I don't know what these systems are doing at every level. I'd rather share that uncertainty with the viewer than paper over it with a wall label.

So I suppose the materiality of the mirror continues the logic of emergence in HEART MANTRAS. Meaning doesn’t reside entirely in the object or entirely in the viewer. The prompt or the output. The “original" or the variation. The “human” or the “machine.” It happens between them. The encounter, the entanglement, is the artwork.

KV: Many discussions about AI focus on intelligence. Your work often seems more interested in behaviour. Do you think society ultimately responds to what a machine is, or to how it behaves?

AR: Society responds to behavior, full stop. We have no direct access to what anything is, including each other. We infer minds from behavior in people, in dogs, and now in machines, using the same wet machinery for all three. Turing understood this seventy-five years ago, which is why his test was behavioral. He swapped an unanswerable question about essence for one you could actually run. I once built a robot that could choose to injure a person, and nobody in the room was thinking about its inner life. They were thinking about what it might do next. So the endless debate about whether AI is "really" intelligent feels to me like it misses where the action is. The social consequences arrive with the behavior, whatever the metaphysics turns out to be.

KV: What role does the viewer play in completing this work? Could it exist in the same way without human presence?

AR: Ghost Currents has no canonical copy. There's no video file and no render farm. The work is computed live, so every encounter is local and all its own. The viewer is more than a witness here, because the piece is quietly watching the room. A camera reads motion, and your movement stirs the fluid. You're inside the weather, whether or not you can prove which swirl was yours. I tuned that influence to sit right at the edge of what you can verify, so participation comes with a little doubt built in. The camera is really just the first control surface. As the work evolves, others may join it, hardware you can actually put your hands on, and each one converts a little of that doubt into certainty, so over time the question shifts from whether you're part of the piece to how much of the weather you can claim as yours. As for whether it could exist without human presence, the system would certainly keep running. If the camera finds nobody, it makes its own weather, and it would do that in an empty room forever without complaint. But I'd argue the work wouldn't survive the vacancy. Half the piece is the computation and the other half is a person finding pattern and intention in it. I count on that attribution the way a painter counts on light. If a painting hangs in a museum and nobody is around to see it, is it still art?

KV: The exhibition begins with Conway's simple rules and concludes with artists exploring language, agency and collective intelligence. Looking ahead, what do you think will be the next emergent property of computational systems that artists should pay attention to?

AR: I think the next thing worth watching is emergence between systems. We're heading into a world with enormous numbers of machine agents interacting with each other at machine speed, negotiating and trading with each other. Let two models talk long enough and they drift into shorthand nobody taught them. Scale that up and you get something like machine culture: conventions, dialects, maybe even aesthetics that exist between machines, with humans in the anthropologist's seat rather than the author's. The other one, closer to home for me, is computation reaching into matter. Systems that don't stop at generating images but negotiate their way into fabrication and physical objects, so that software output becomes something you can hold in your hands.

Artist Profile

Next: Collector’s Preview

— 4th September