TOrn between screens and streams
I grew up in a small, kooky town in the southwest of Devon, called Totnes. Totnes is known for a lot of things, some good, some not so good, but there are two things I love about my hometown.
The first is the town's historical commitment to the environment: the early agrarian reform of Dorothy and Leonard Elmhirst in the mid-1920s, the Dartington Amenity Research Trust in the late 1960s, Schumacher College in the 1990s, and, most recently, the transition towns movement of the early 2000s. These legacies had a profound effect on me growing up, and my education reflected it. At secondary school, I spent a whole term on climate imaginaries - my first introduction to STS thinking - presenting the news from a future above 1.5 degrees, I stood up in class to tell everyone what Coca-Cola was doing to rivers in India, I gave up meat, then dairy, I marched with youth climate movements, wrote letters to local councillors about climate breakdown, and refused, on principle, to get in a car if I could walk or cycle instead.
The second thing I love about Totnes is its history of computing. Charles Babbage went to King Edward VI Grammar School, the same school I went to. The first computer I ever touched was an off-white Dell Dimension, given to my mother for free as a mature student with dyslexia. I was the youngest in the house but the one always called when it broke. I learned early that switching it off and on again fixed most things. I was no Arron Schwartz but as a teenager I had picked up some basic programming (mostly so I could make cool themes for my Tumblr blog) and understood that the terminal was a place where I could make things happen. That felt like magic, and it stayed with me. It took me out of Totnes, first to Manchester and then to London, for a computer science degree and then a job at a big tech company.
I was good at that job. But my worry for the environment outgrew my love of code. So I left that job, much to my mother's dismay, and instead returned to education focusing on environmental data science. A few years later, I was back in Totnes as I received a PhD position at the University of Exeter. The PhD was in Environmental Intelligence, a new interdisciplinary practice that aims to reconceptualise AI through a pedagogical programme that asks doctoral candidates to hold data science and a humanities discipline in the same hand, so you actually understand where your data comes from and what it really costs to collect it. It truly felt like coming home to study the two things that Totnes had taught me to love from such a young age.
We must "harness" AI, or "exploit" it, as if the technology were a wild dominatrix waiting to be tamed.
Yet something troubled me in this homecoming. For I quickly noticed something jarring about how people talked about AI and its role for dealing with the environmental crisis.
On one side were people who believed, optimistically, that AI will somehow be a force for good. AI will give us new tools to monitor and cut emissions. When describing this view, folk often reached for strange, almost sadomasochistic language. We must "harness" AI, or "exploit" it, as if the technology were a wild dominatrix waiting to be tamed. On the other side were people who opposed AI outright, and treated anything to do with it as nothing more than late-stage capitalism, fascist machinery, and something to resist at first sight.
Neither position sits right with me. I am neither a total optimist nor a total pessimist. While I understand that all forms of life are losing their homes, not only to floods and fires but to the technological systems built to tackle them, systems that watch us closer and closer, repeat old colonial patterns under new technoscience names, and wear down trust of the oldest and hardiest institutions humans have built. At the same time, I see that computational infrastructures are already deeply embedded in how we understand the environment itself. Evolution, the diversity of life, even time itself, are all concepts we know largely through data and models built to hold more knowledge than any mind could carry. And you cannot simply switch that off and on again.
Instead of picking a side, I follow my supervisor, Sabina Leonelli, in thinking about environmental intelligence through five ethical premises. First, intelligence is distributed, embodied and situated, not something that sits alone in a machine or a mind. Secondly, human judgement is essential to deciding what counts as evidence in the first place. Thirdly, trying to control the natural environment tends to undo itself. Fourthly, we are in an urgent socio-environmental crisis that cannot be quietly worked around. Finally, entrenched inequities carry a heavy environmental and social cost that someone, somewhere, is always paying.
AI is neither saviour nor villain. It is a set of practices, built by particular people, funded by particular interests, and put to particular uses.
These premises now shape my ethical view on AI and help resolve the tension I first felt as a child in Totnes, torn between screens and streams. AI is neither saviour nor villain. It is a set of practices, built by particular people, funded by particular interests, and put to particular uses. If we want AI and education to serve environmental ends, we need to ask who is doing the harnessing, who is being exploited, and whose home is on the line? Environmental Intelligence does this by asking us to stay close to the specific and the situated, rather than reach for grand narratives in either direction. That feels like the more honest place to think from, and to teach from, when it comes to AI, ethics and education today.
