"How Do We Break Free From Fixed Bodies and a Fixed Society?" One Artist Answers With an AI Performance

He fuses performance with artificial intelligence, hoping to challenge what we think of as "normal."

Born in 1984, Marco Donnarumma is an artist and scholar who holds a PhD from Goldsmiths, University of London, in performance art, computing, and the theory of the body.

As an artist, he has always kept the relationship between technology and the body at the center of his work. On one hand, he's interested in how the body exists under the shaping force of society and culture; on the other, in how technology, as part of that culture, in turn shapes the body.

From April 19 to 20, 2019, ALIA: ZŬ TÀI (《异体：组态》), written and performed by Donnarumma, was staged at the Goethe-Institut in Beijing. The piece brings together Donnarumma, dancer Wu Yandan, actress Chen Lingling, and three "prosthetics with a will of their own," and asks: what kind of identity can artificial intelligence and robots produce? How do these technologies shape the way we imagine the human body and consciousness?

The two "self-aware prosthetics" in the performance are shaped like the bones of a human limb — AI robots that Donnarumma designed and built himself. They can sense different forces through their sensors and respond according to pre-written programs.

Before the show starts, the stage is nearly bare — two computer monitors and the prosthetics sit downstage. Once the performers enter, one of them types a line of code on a keyboard, and the screens immediately start generating a scroll of shifting data as the prosthetics begin to writhe. ALIA: ZŬ TÀI has no dialogue at all; the three performers enter at different points in the show, and their bodies engage the prosthetics in relationships that swing between control and resistance.

In a post-show talk, Donnarumma gave a brief rundown of how the AI robot works. He fitted four motors at different joints of the prosthetic, each running its own distinct algorithm, each capable of responding differently to what the sensors pick up. Each motor stands in for a different neural network, controlling the movement of a different part of the robot. Donnarumma is the one who programs how the robot responds, but exactly when it will make what kind of judgment call is unknown even to him — so in that sense, the speed and direction of the robot's movement are decisions "it" makes on its own.

Click here to watch a clip from ALIA: ZŬ TÀI

ALIA: ZŬ TÀI is one piece in the 7 Configurations project; ALIA, in the title, is Latin for "the other." Donnarumma describes the piece as a "rite of initiation" — the performers on stage try, by every means available, to connect with a robot so unlike themselves, and in that process, somewhere between merging with it and resisting it, the robot ends up governing the performers' behavior, even reshaping how they understand the stage itself.

"When the robot moves, it's almost impossible to resist imagining that it's alive — we even have to reach for the imagination to interact with it at all," Donnarumma says, calling this one of the most important things ALIA: ZŬ TÀI reveals once it's performed. "We project our own problems and emotions onto the machine. The struggle we feel with the robot is really a struggle with our own desires. Which means that when we talk about robots, we're ultimately still talking about ourselves."

Donnarumma joined a band as a bassist at 15 and started playing live shows. At 16 he got his first computer and started making his own electronic music as a party DJ and producer. After seven years of making music this way, he began to feel electronic music's limits when it came to a live, physical presence, and started to doubt whether music built out of such fixed, formulaic electronic signals could really express thought or feeling. That doubt pushed his practice toward performance — he wanted to use his own body on stage to explore ways of making electronic music more expressive.

In 2010, Donnarumma built a device called XTH SENSE. Worn on a performer's body, it picks up the sounds of the heart, blood, and muscle, then amplifies and transforms them. A performer wearing XTH SENSE can control their own bodily responses to improvise music live, in the moment.

XTH SENSE won first prize in the 2012 Guthman Musical Instrument Competition, run by Georgia Tech's Center for Music Technology. Donnarumma calls this style of music — built on a feedback loop between biology and sensors — biophysical music, a form that dissolves the boundary between the biophysical and the electronic. As Donnarumma put it, "The essence of XTH SENSE may be that it gives performers a more direct voice from inside their own bodies."

In 2014, Donnarumma began 7 Configurations, a project that is a series of works combining performance and installation art. By 2019 he had completed five pieces in the series: CORPUS NIL, AMYGDALA, EINGEWEIDE, ALIA: ZŬ TÀI, and CALYX 5.

Besides ALIA: ZŬ TÀI, both CORPUS NIL and EINGEWEIDE are performance works as well.

A clip from CORPUS NIL

CORPUS NIL is performed by Donnarumma alongside an AI machine. During the performance, the machine analyzes physiological changes picked up by sensors worn on Donnarumma's body, and uses those changes to shape elements like the lighting and the score in real time — and Donnarumma, in turn, adjusts his own performance based on what's happening on stage.

Click here to watch a clip from EINGEWEIDE

In EINGEWEIDE, Donnarumma and another performer achieve a full fusion of AI and body. Donnarumma wears the "self-aware" prosthetic over his own face, which blocks his vision entirely — from that point on, he adjusts his performance and his movements based purely on how the robot responds.

AMYGDALA and CALYX 5 are two installation works.

AMYGDALA is also the name of an AI robot whose task is to continually cut and trim a "skin" — and, through that cutting, keep learning how to handle new situations. The "skin" is actually a distinctive fabric Donnarumma created himself, one that gradually stiffens over time. The AMYGDALA robot is also fitted with sensors, and it changes its behavior based on the shape and condition of the material it senses — the piece is mainly a demonstration of how an AI robot learns.

CALYX 5, meanwhile, exhibits the cut and trimmed "skin" itself — a record of the AMYGDALA robot's learning process.

Across the 7 Configurations series, Donnarumma deliberately foregrounds the relationship between the body and technology, going so far as to create something like an "unknown creature," suspended somewhere between human and artificial intelligence.

Before explaining the work, Donnarumma offers two examples to unpack what "normal" even means.

He starts with a book from Western history that was once considered "science," full of purported "knowledge" about female reproduction. One detail: if a pregnant woman had a nightmare, the book claimed, she would give birth to a monster covered in hair. "It wouldn't matter if nobody read this book," Donnarumma says, "except that for close to two hundred years, people believed Aristotle had written it, and did a great deal of scientific research on that basis."

The other example is a photo of a "female robot" developed by Japanese scientists. "It's a beautiful female figure — her face looks perfect, like she's had plastic surgery. When a scientist wants to show the rest of the world a female robot, this is the design he chooses. Why is that?"

Donnarumma believes every body is different, and he wants people to understand that the human body is a mixture that is always growing and always incomplete at once — not some "normal" concept handed down by a form of power. That's why his work creates the figure and relationship of a "different body": he wants audiences to look past the idea of "normal" when they encounter that "different body," and from there to imagine "how we might break free from fixed bodies and a fixed society."

The following is an edited transcript of the interview:

Q = questions from the audience and reporters

D = Marco Donnarumma

Q: What's the function of the two screens in the performance? As an audience member, I couldn't actually read them, but they seemed to carry some kind of concrete symbolic information tied to the rhythm of the robot's movement.

D: Those two screens are meant to make the audience feel confused and disoriented. All the performer does is type in a line of code to set the robot moving, but that code was written well in advance. What you're watching is really the process of how we configured the software — every time the robot moves, some letters appear on screen, and the whole screen fills up with code that can't actually be read. That's the process of how we developed this technology. We want the audience to get a sense of how these machines get built.

Q: In the program notes for ALIA: ZŬ TÀI, you mention the idea of "self-aware prosthetics." How should we understand this self-awareness? Is it the same as human self-awareness?

D: It does have a kind of "self-awareness," but it's not the same as ours. The robot's self-awareness means it can sense the parts that make up its own "body" — if you push it or pull it, it can feel that force. But unlike human self-awareness, it has no idea it's on a stage, no idea there's an audience out there.

That's also what makes it interesting — in that sense, it has a degree of understanding, the way humans have a kind of agency. What separates the robot's agency from a human's is exactly what I'm curious about. An AI robot's agency comes mainly from its algorithms, but it also has some capacity, some potential, for improvisation.

Q: The improvised responses the AI robot makes on stage still look fairly "safe" — you can't predict the exact changes, but its reactions never seem to go too far outside expectations. Did you ever consider making the robot's responses less controllable when you were designing it?

D: Part of it is a limitation of the robot itself — it's just the shape of a limb, it can't run, it can't jump. That's also a choice we made deliberately: our budget is limited, and a bigger robot is more fun, but also a lot more expensive. There's another reason too — the more we keep the robot's movement inside a "safe range," the fewer accidents can happen on stage.

But we do want more room for improvisation. More improvisation means more freedom, but also more risk. When we genuinely can't predict what the machine is going to do, we end up startled by it ourselves — and that's interesting, for us and for the audience alike.

Beyond that, I've been doing this research and this kind of work for almost four years now, and I still feel like I've only scratched the surface of this technology. There's a lot more still to discover.

Q: Is there a way to shift perspective and think about the relationship between machines and humans without putting humans at the center?

D: I think there are a few layers to this. First, we need to break down certain boundaries — that's both the most important and the hardest part. There are several boundaries at play: the boundary of the stage, and the boundary of performance art itself. How do we break those down? We actually tried this in our piece — the three performers come from different backgrounds. I come from body art, Wu Yandan is a choreographer, Chen Lingling is an actress who also works with puppetry. We can pool our backgrounds and figure out, together, how to make the best possible work.

Then there's the technical layer. It's important to step outside habitual ways of thinking, which is why we designed the piece the way we did. Something interesting: most people don't realize that a lot of research into this kind of robotics and AI actually comes out of the meat-processing industry and the auto industry. Most of this research is really aimed at building AI and robots for those two industries. Our funding comes from those industries too — can you imagine that?

Sometimes we can also draw on robots used in other industries that have some AI built into them. We don't necessarily agree with how those robots work, but they still bring us different backgrounds and different possibilities. In our piece, you can see that some of what we're imagining isn't tightly bound to robots or AI at all — but I think that's also a good way of leaping outside the traditional approach.

Q: In your explanation of the work you mention how important it is to break down boundaries. Even though the performers come from three different artistic fields, ALIA: ZŬ TÀI is still, at bottom, a body-performance piece. How does the way you look at the body differ from how the other performers look at it?

D: A professional dancer might have started body training at age six, and kept training throughout their entire dance career. On one hand, that gives a dancer an intimate knowledge of their own body; on the other, it also locks that knowledge into a fixed pattern. When a professional dancer is dancing, part of their mind is still counting beats; when they choreograph, they're working from the score, so they tend to pay more attention to sound or musical content.

When I use my body, I'm more focused on what changes it goes through under what conditions, on how far I can push the body toward some limit. Treating the body as a kind of experimental site makes the whole thing more fluid.

Q: Why does "the body" keep showing up as a theme in your work? Why does the body matter so much to you?

D: First, because I've found that every one of our bodies is complex and fragile, and no two bodies are ever exactly the same — I find that fascinating. Second, because the way society understands and interprets the body directly shapes how we survive within it. So I pay attention to how society understands our bodies, what questions we ask about our own bodies, and which questions we simply never ask.

There's one question I always want my work to answer: how does technology shape the way we understand the body? In the age of the smartphone, for instance, we've learned how to unlock a phone with a gesture; when people used to talk about the act of "mining coal," most people would mime holding a tool, but now they might just as easily mime operating a machine.

How we use technology — does technology shape a "normal" body? Do we use technology in order to be shaped into a normal body? But there is no such thing as a normal body to begin with. Every body is different, and every body is incomplete in its own way; most of the time, what we're really doing is bearing and accepting that fact. That's what I want to explore.

Q: I noticed the fusion between the performers and the AI never goes smoothly — there are various attempts, and relationships of both rejection and control — so I sense a critical stance in your work. But based on how you've explained the piece, it sounds like you're not critiquing technology itself, but a fixed idea of "normal." Is that an accurate reading?

D: That reading is mostly correct. I think the concept of "normal" is something power imposes on every one of us, and technology is, in fact, a form of power — a medium through which "normal" gets imposed on people. I don't personally take a critical stance toward technology itself. Humans are never going to stop inventing new technology; we've been creating technology continuously since our ape ancestors, and we'll keep innovating in the future.

So technology itself isn't bad — how it gets used is a different matter. And the power of technology genuinely goes far beyond what we imagine. I don't mean that decades from now robots will gain intelligence and start conquering humanity. What I mean by "powerful" is the terrifying scale of the unintended consequences that come out of how we use it — consequences that go far beyond what we predicted. Climate change right now, mass animal extinction — these are side effects of the very technologies we're already using.

Q: What's the relationship between your background in the sciences and your artistic practice? Does knowing this much about technology fuel your imagination, or does it also become a kind of constraint?

D: In some ways it makes me more imaginative — I tend to think more deeply about technology than most artists do, and I understand what's behind how it's presented and used more thoroughly, which makes me more critical of it too. That also means I keep getting research funding, but it's not always easy to use that funding to make the kind of work I actually want to make.

Cover image and images throughout courtesy of Marco Donnarumma.