Environmental machine learning as artistic research practice.

Posted Oct. 12, 2018 by Theun Karelse

Environmental machine learning as artistic research practice: how does such a mix of ecology, technology and art make sense in today’s world?

The field of environmental machine learning caught my attention about eight months ago, when I became acquainted with the work of Theun Karelse. I was looking for a way to continue on the topic of climate change, as I had been studying sustainable development and ecological anthropology. I had learned that some problematic ideas at the roots of the anthropogenic damage to earth are individualist perspectives of self-reliance or ecological independency, anthropocentrism, and the idea that some entity called ‘Nature’ is separate from humans.

In this human-centered view, Nature is merely a set of resources to exploit, or even force to fight against and exterminate. For centuries, it has lead us to think, build and behave as if humans are central in all domains of existence (Morton 2018). This human-centered and individualist worldview maintains a sense of exceptionalism in which humans are substantially different from or superior to other beings - with all unintended consequences thereof (ibid.; Haraway et al. 2016). Language matters in forming worldviews. For example, the concept ‘sustainability’ is risky, as it is all too often anthropocentrically scaled or used to cover up destructive human practices (see Morton 2018; Tsing 2017). That’s why terms such as ‘multispecies ongoingness’ or ‘multispecies collaboration’ (Haraway 2018), and ‘multispecies resurgence’ (Tsing 2017) seem more fitting. These alternatives for sustainability go beyond the human-centered focus and acknowledge the alignment of humans to multispecies dynamics.

And yet, whether you call it the Anthropocene or not, here we are - living in a time in which human impact on earth is causing major imbalances in ecological flows. The urgencies of the Anthropocene are also very much nonhuman urgencies, such as the current sixth mass extinction event. And vice versa: nonhuman urgencies directly or indirectly cause all kinds of urgencies for humans. In Timothy Morton’s words, to oppose anthropocentrism is to understand our inclusion in the earth’s ecology, ‘as one being among others’ (2018). The realisation of these earthly problems also lead to a feeling of responsibility; human agency has done a lot of damage, now it is our turn to use human agency to be more capable of response – or in Haraway’s words to be ‘response-able’ (Latour 2014; Haraway 2016).

Realising it would be impossible to solve this complex tangle of earthly problems, I started to search. Not for more numbers on species extinction or maximum degrees of global warming, not for ready-made solutions either - I strongly doubt they exist. The search was aimed at something Timothy Morton and Donna Haraway aspire as well; a way to grasp this earthly problem that we share with all other critters, to be truly present. And, at the same time, to look for a non-anthropocentric, vulnerable, activist stance amidst the trouble that sparks some hope and imagination on more sustainable earthly symbiosis. I found three projects that somehow hold this view: terra0, Dark Ecology and Random Forests. Here, I will elaborate on the latter one.

Random Forests is an adventurous project that explores the limits, but overall the opportunities of environmental machine learning. It is an investigation of how today’s most advanced technologies can be more inclusive of ecological processes, while acknowledging that technology is always inextricably entangled within complex systems. Innovative yet history-sensitive, it studies how carefully designed machines (as always-works-in-progress) collaborate with, learn from, and communicate intimately with ecological symbionts. Random Forests sparked my interest because before I barely thought beyond the dichotomy of artificiality and wilderness, nor had I imagined what a more ecological approach integrated into technology could look like. This project opened a whole world of environmental self-learning machines, nonhuman agents, speculative scenarios, and novel ecosystems that I had never before encountered. What really appealed to me about Random Forests is the invigorating quality of its approach. The approach of the project is not to stress how bad things have gotten. It does not carry out the painful, guilt-causing or quantitative character of many of the climate studies related works of today, while it does remind one of the necessity to face the current earthly situation caused by human activity. Its approach is activating, energizing and sparks imagination.

I have been fortunate to join Theun during part of his fieldwork sessions and interviews for Random Forests. His enthusiasm for the topic was contagious; most of what we learned is based on constant curiosity. It has been a transformative experience in that it has strongly shaped my Master’s research as well as my personal perspective on the world. For example, we speculated and set out a concrete imagination of an environmental AI on Terschelling, tasked with the conservation of the UNESCO area. The AI-speculations were based on site-specific research (walking and cycling on Terschelling, speaking with residents and listening to histories of the island), and interdisciplinary discussions until late in the evening. What resulted was a machine that can be wayward and quirky, sometimes unsettling. We wondered, for instance, if such a machine could ever be able to ‘think outside the box’, or make exceptions to programmed rules. We gave it a set of sensory skills through speculatively installing sensor-networks in the landscape. We figured that certain species, such as goose or lichens, could be interesting collaborators for the AI. Unexpected combinations of engineering, art, design, philosophical wondering and unlearning of human categories kept surprising me throughout the research.

To unlearn human categories and thus to think beyond anthropocentrism means ‘to question the tissues of one’s knowing and ways of knowing’ (Haraway 2016, 122). It invites one to ‘become playful about the lack of an obvious solid ground of meaning, one obvious scale on which to see and act’ (Morton 2018, 211-2). A playful method that recurs in artistic research projects I studied is speculative culture. Speculative culture is about the possible, asking ‘what if…’ - exceeding the limits of what is ‘commonly known’ or ‘makes most sense’ (Dunne and Raby 2014). Rather than simply describing or maintaining reality, speculative culture is concerned with changing it. Building scenarios about the path between the present and possible futures exposes all kinds of obstacles in the process. Thinking about what could happen instead of what should happen prevents speculative scenarios from becoming didactic or moralistic (Dunne and Raby 2014). It is a playful but serious method, that helps to think beyond taken-for-grantedness. Speculation helps to hybridize nonhuman entities and explore new research tools, effectively highlighting urgent issues without suggesting a ‘better’ way to deal with them.

What I have learned about speculative culture from participating in Random Forests is that there is always one or several scenarios to be worked out; there are contributor-participants from different backgrounds; the speculation is not generalized but site-specific; the specificity of the session is a powerful tool to reveal specific societal and environmental issues that could get in the way - it is, in short, a non-reductionist method because it takes context specific tensions and frictions into consideration. Furthermore, humour plays an important role in the speculation to come to new imaginations and ideas; and speculation is an ongoing process - people will not stop speculating, asking questions and imagining on the topic even after the workshop is closed.

Another quality of artistic projects like Random Forests is practice-based research. Where social sciences mainly stick to the thinking part, the artists I’ve met are thinker/makers - a term borrowed from Haraway meaning ‘those engaged in the inextricable thinking/making practices called art’ (2016, 89n75). The strength of making, here, lies in the openness of the creative practice; it could end up anywhere, things work out as the process of thinking/making goes along. The artistic research projects are the locus of speculation, but also of the materializing of the imaginations that come out of speculative culture. Speculation is creative in the sense of creating something new in mind (like new words, theories, concepts, perspectives, agencies, questions, or problems), but also in matter (like new artworks, software, robots, organisms, or ecologies).

Haraway proposes a way of thinking and acting beyond individualism with the word ‘sympoiesis’, which means ‘making-with’ or ‘collectively-producing systems that do not have self-defined spatial or temporal boundaries’ (2016, 35-6; 58). It describes a commitment to collaboration of all different beings on earth, as we are amidst urgencies that are not just human urgencies (ibid.). As opposed to autopoiesis, which means that systems, organisms, persons, things can be self-constitutive and self-making, sympoiesis implies that ‘earthlings are never alone’ (Haraway 2016, 58, emphasis hers). This making-with is always done together with all kinds of beings who can be called companion species. Building from the Latin cum panis (‘with bread’) Haraway emphasises how much we share with our ‘ontologically heterogeneous partners’ in the ecological assemblage (ibid., 2016).

Having said that, what to do with this new category of intelligent artificial nonhumans? Could we regard this new machinic species as companion species? This is a main question investigated in Random Forests through speculative culture by thinker/makers. At a fast pace, machines are designed to do operations that formerly required human practice. As anthropologist Tim Ingold observes, these human practices are now reshaped in their interactions with these technologies. Rather than solely being replaced or surrounded by these machines, humans are involved in new encounters and collaboration with these machines (2011). Similarly, the machines I encountered in this research are not going to ‘take over’ nonhuman activity, or fully replace organic elements or processes of complex ecological systems for artificial ones. Neither are they forms of geoengineering or ecomodernism. More accurately, they are designed to become new symbionts, making the environment with their context-specific companion species. Operating from a deeper ecological understanding than conventional machines, these technologies can actively contribute to multispecies resurgence. Taking the artistic approach seriously and including it in engineering and large-scale socio-economic decisions would create a more sensible application of autonomous machines in daily life, a more response-able one.

The Random Forests project has shown me that it is possible to remain optimistic about the current and future state of this planet, while keeping a critical stance. I think it is mainly the ability to re-imagine what was taken for granted that can save us from pessimistic apocalypse-thinking as well as overoptimistic neglects of ecological and social urgencies. So, after encountering this inspiring project, what to do with it? Take it as a method or a tool to think and act in the world - as active, response-able part of the world. Or take it as an artistic endeavour, an exhibition in the mind to wander through, as a poetic view on earthly collaboration between species. See it as an engineering challenge or an eye-opener. In whichever way you experience Random Forests, it bends and twists what we know, shows us the usually overlooked. It helps recognizing the world from a more-than-human perspective, and reminds of interdependencies across species. This, it seems to me, is the important first step towards response-ability for multispecies ongoingness. Challenges abound, I agree. But that does not need to stop us from trying and enter an exciting perception of the world that we share with so many inspiring others. It would be of great value to academic research, public societal debates and educational programmes to integrate artistic research methods.

Altogether, this is a call for openness in thinking, for taking seriously those experiences that spark imaginations, for curiosity instead of fear about the unknown and unfamiliar, for art-science collaborations as well as human-nonhuman sympoietics. It is inspired by artistic research projects that - in the broadest sense - rethink and reimagine what it means to live in a time called Anthropocene.


Terschelling dune landscape - During the speculative sessions, I learned to look with a lot more imagination to such landscapes, and all of a sudden sensor networks and ecological agents appeared. Random Forests’ perspective offers space to notice how plants, soil microbes, animals, and weather patterns as ecosystem symbionts have agency as much as humans do.