Jenna Sutela On Machine Learning and Interspecies Communication

The artist discusses her new piece on computer- and bacteria-generated language

Based in Berlin, Finnish artist Jenna Sutela's installations, texts, and performances seek to identify and react to precarious social and material moments, often in relation to technology. Most recently, the artist has been exploring exceedingly complex biological and computational systems, ultimately unknowable and in a constant state of becoming.

In early 2018, Sutela set about exploring this further after joining n-Dimensions, Google Arts & Culture’s artist-in-residence program at Somerset House Studios. The opportunity led her to collaborate with fellow artist Memo Akten and work with engineers at Google Arts & Culture in Paris to investigate whether computers could channel spirits instead of algorithms.

The result is nimiia cétiï, an audio-visual work inspired by experiments in interspecies communication that aspires to connect with a world beyond our consciousness. The piece uses machine learning to generate a new form of communication by documenting the interaction between a neural network, audio recordings of early Martian language and footage of the movements of space bacteria.

Here Sutela explains more about her four-month residency and the work she’s created, as well as providing a little more insight into her practice as a whole.

How would you describe your work to someone unfamiliar with it?

I often work with words, sounds, and other living materials, exploring biological and computational systems. For example, I’ve collaborated with Physarum polycephalum, the single-celled yet "many-headed," yellow species of slime mold, also known as a natural computer. My recent project Orgs focused on decentralized organisms and organizations; the possibility that decentralist eco-centric technology might take over, reclaiming its pre-human domination.

A view to the kombucha heart of Jenna Sutela's homebrew computer, Gut-Machine Poetry, 2017, photo: Mikko Gaestel (From the collection of Somerset House)

Can you describe your piece nimiia cétiï created as part of your residency? What was the inspiration behind it?


nimiia cétiï is an audio-visual piece using machine learning to generate a new written and spoken language based on a computer's interpretation of a Martian tongue from the late 1800s, originally channeled by the French medium Hélène Smith and now voiced by me, as well as the movement of Bacillus subtilis, an extremophilic bacterium that, according to recent spaceflight experimentation, can survive on Mars. This bacterium, by the way, is also used in the making of nattō, or fermented soybeans, a classic Japanese probiotic considered as a secret to long life.


The machine, in this work, is a medium, channeling messages from entities that usually cannot speak. But the work is also about intelligent machines as aliens of our creation. There's an interesting link between the project of talking with aliens and the problem of talking with machines. We built (at least some of) the aliens ourselves and now the challenge is to understand the nonhuman condition of these machines that work as our interlocutors and infrastructure.


Visually what can we expect from the piece?


A video shows a computer watching footage of the Bacillus subtilis bacteria under a microscope and generating a script, or calligraphy based on an analysis of what it sees. Imagine a pen suspended from a long piece of string, resting on paper that's slowly sliding sideways. Raw force from the movements of the bacteria knocks the pen around, leaving marks on the paper. Beyond this still rather anthropomorphic approach, there's also a brain-like illustration, or a map of what's happening in the bacterial video from the computer's perspective—the logic of which is inexplicable to a human—as well as the computer dreaming up bacterial movements using a future prediction algorithm as a source for optical flow.


Audio interacts with the bacterial movements. What you can hear is the computer reorganizing or mimicking the early Martian language. A network trained on my voice looks at each frame of the video and produces a short block of sound that it thinks matches that frame, or the configuration of bacteria in it.

Special edition of Orgs: From Slime Mold to Silicon Valley and Beyond, a publication edited by Jenna Sutela, Garret Publications, 2017 (From the collection of Somerset House)

How does the piece relate to your work as a whole?

Working with text, or narratives on the one hand and microorganisms on the other, I'm constantly challenged by communication, trying to find a means beyond language for interspecies collaboration and then ways to express it linguistically.

nimiia cétiï continues in the footsteps of Gut-Machine Poetry, a project where I introduced entropic processes into computing via inserting fermenting foodstuff into the guts of a computer. A few years ago, together with collaborators Vincent de Belleval and Johanna Lundberg, we created a microbial/computational poetry culture: a “wetware random number generator” based on microscopic footage of a symbiotic colony of bacteria and yeast in a kombucha tea ferment connected to a series of letters. The stochastic movement of yeast eating sugar affected the jumbling of letters on a live website. The letters were based on my writings about code laws and organic behavior, the gut-brain connection, etc.

Your work often relates to technology in some way, what draws you toward this subject? And how does it manifest itself in your work?

Like some of my favorite science fiction, I believe art has the ability to generate a more advanced image or experience of now and the emerging next phase before it comes to existence in other registers of life. I try to engage with futuristic and also ancient materials in a non-linear way, while creating spaces and devices that may carry us to parallel worlds.

Jenna Sutela's From Hierarchy to Holarchy, 2015 shows a living slime mold drawing itself across an organizational maze. Photo: Mikko Gaestel (From the collection of Somerset House)

What have you learnt during your residency?


I've learned about the possibilities that machine learning opens and the challenges it poses to communication and expression. I've learned that computers should be approached on their own terms, not anthropomorphizing them too much. For example, the so-called “black box problem” in machine learning means that it's sometimes hard to explain how an AI has come to its conclusions.


What did you enjoy about collaborating with the Lab’s engineers?


It was exceptional to have Head of Innovation, Damien Henry from Google Arts & Culture and artist resident Memo Akten from the Somerset House Studios, these two brilliant programmers and artists to work with for such a substantial period of time. They responded to and challenged my ideas while walking me through the depths of machine learning and realizing things that I could only dream of technically speaking.

Physarum polycephalum, or the “many-headed” slime mold ingested as a form of artificial intelligence in Jenna Sutela’s performance, Many-Headed Reading, 2016, photo: Mikko Gaestel (From the collection of Somerset House)

Have there been any challenges during your residency?

I was originally interested in errors in machine translation. I wanted to use the opportunity of doing a residency supported by Google Arts & Culture to get access to data showing the Google Translator making mistakes and thus revealing something about the computer's mechanisms of cognition. I've been inspired by the work of cognitive scientist Douglas Hofstadter who researches slips of the tongue and speech errors in human language for the same reason. However, to my disappointment, it was not so easy to access this sort of data...

What will you take forward in your practice after this experience?

I feel like the residency was only the beginning of a longer project on machine learning and interspecies communication. Ultimately the aim is to contribute to the development of a culture based on symbiosis rather than the survival of the fittest narrative—organic and synthetic life forms included.