Cybernetic Studio

The Cybernetic Studio is a creative research space at the Australian National University’s School of Cybernetics. Here, artists, engineers, computer scientists, environmentalists, researchers, and organisations come together to make things—and break things—to explore how technology can be designed and built for safe, sustainable futures.

From the beginning of Cybernetics, material practice has been a key part of knowledge discovery. There are heaps of examples–Wiener’s hearing glove, Grey Walter’s tortoises, Ashby’s homeostat, Beer and Pask’s fungoid ‘whisper’ systems, and many more.

Welcome to the Cybernetic Studio—where creativity, computation, and curiosity converge.

Cybernetics is about steering—navigating complex systems of people, technologies, and environments. In today’s world of rapid technological change, that means helping humanity chart a course toward safe, sustainable, and equitable futures.

The Cybernetic Studio is our experimental space for exploring what responsible innovation really looks like—a place where we prototype ideas, test assumptions, and make the invisible visible.

Our Human-Scale AI philosophy invites us to slow down and look closer: to make the workings of computation tangible, perceptible, and open to reflection. By bringing technology within human reach, we empower people to shape it with care and intention.

Through collaboration with partners across industry, government, and the creative sector, we’re developing artefacts, frameworks, and practices that help organisations design for safety, sustainability, and accountability.

Together, we can build technologies—and futures—that keep humanity, and the planet at the centre.

Dr Ben Swift

Human-Scale AI is the theme for the Studio’s 2025/2026 activities. Humans have been making computational devices for thousands of years. The Antikythera mechanism, abacuses, slide rules, and the Difference Engine. And in the mid-20th century we get digital computers, and the whole “computation” thing really takes off. I’m not gonna do the “pull smartphone out of pocket and compare it to the size of the computers of yesteryear” bit, but not because it’s not staggering—it’s just that you’ve all heard it before and we’re no longer amazed by the computational power we carry in our pockets.

Then in 2022 ChatGPT and Large Language Models break into the mainstream, and we can not only talk to our computing devices—but they can answer Back. Again, this is fundamentally just more computation, it’s just that it speaks our language (because that’s what we trained it to do).

But there’s a cost to the march of acceleration and miniaturisation in computing ability. We can no longer look at a computing device in action and see what’s going on… not even a little bit. We’re increasingly disconnected in our understanding of how computation works.

That’s the background for the Cybernetic Studio’s Human-Scale AI project. We interact with computational devices more than ever before, but we’re alienated from the nuts and bolts of their operation. Our claim—grounded in a history of hands-on experimentation and knowledge creation in the cybernetic “mangle of practice”—is that it is possible to slow these computations down and “blow them up” so that we can see them at work. Human sense perception is amazing—so if we can bring AI within that range; make it a human scale—then we can apply all our sensory tools to try and make sense of it.

Does it mean that we’ll see the matrix, ones and zeroes in neon-green-on-black? Well, no, but that’s not the point. Because even glimpse of computation at human scale helps us better orient ourselves to the way it flows around and through us even when we can’t see it. By making and using these human-scale AI systems we can make better decisions about how to participate, when to abstain, and how to ensure that computation continues to benefit us all at every scale.

Sample Exhibit: The Perceptron Apparatus#

The Perceptron Apparatus is a large-scale interactive sculpture — a reimagining of how early neural networks might be rebuilt if their knowledge were lost. By making the hidden processes of artificial intelligence visible and tangible, this exhibit invites audiences to see, touch, and question the systems that shape our world. It demonstrates the Studio’s commitment to “human-scale” innovation — technology that can be understood, engaged with, and designed for safety and accountability.

Sample Exhibit: My First Language Model#

My First Language Model is a hands-on analogue activity—a reimagining of how modern AI might be taught if its complexity were stripped away to its core principles.

By making the hidden mathematics of language models visible and tangible through paper grids and dice, this activity invites participants to build, use, and question the systems that generate text in our digital world.

It demonstrates the Studio’s commitment to accessible AI literacy—technology education that can be understood through direct experience, engaged with across ages and backgrounds, and designed to foster critical thinking about AI capabilities and limitations.

Sample Exhibit: PANIC!#

PANIC! is an interactive installation and computational research project—an exploration of how semantic information might degrade and transform when generative AI systems are connected in feedback loops, recursively feeding outputs back into inputs.

Through these time-lapse visualisations we’re invited to trace, measure, and question how semantic content drifts through repeated AI transformations. The video on display loops through different trajectories, showing how the same system produces different emergent behaviours depending on initial conditions: the starting prompt, which generative AI models are connected together, and the random seed.

It demonstrates the Studio’s commitment to empirical AI research—studying complex information-processing systems through controlled experimentation, mathematical analysis, and open scientific practices (all code and data is available on GitHub) that enable reproducibility and further inquiry.