EVA KOZANECKA

The Garden Project

Rewilding the planet, one garden at a time 

Status: In development

Speculative Fiction   

"X, a software engineer and Y, a filmmaker were kicking up dust recently at Tempelhofer Feld, a former airport and now, public park in Berlin, Germany, trying to determine the feasibility of planting a garden in the dry and sandy plain. 

The soil and lawns were parched after an unusually dry summer spell, but the soil’s sand-like quality was the result of a much older natural phenomenon: a retreating glacier in the Little Ice Age– a distinct cool period between the Middle Ages and the first half of the nineteenth century– that left behind what archaeologists call: a sand-bar. It stretches across most of Northern Europe, and is one of the reasons why community gardens that have since sprung up in the Tempelhofer fields use raised boxes for plantings.

For X and Y, however, the question at the heart of this visit was not if a garden could be sustained in this location, but by whom and with what tools? 

Since meeting five years ago, the two have been co-leading a program at Google that pairs artists and engineers together to realize projects using machine intelligence. The program–titled, Artists + Machine Intelligence– is part of a research design group that believes machine intelligence can (and should) augment human ability. The goal of each artist and engineer collaboration is to try to answer exactly how?

Nature may not need help from humans, never mind technology. Earth’s unprecedented warming over the last century is overwhelmingly attributed to humans putting too much carbon in the atmosphere. But in the US, for example, municipalities are running short of scientists trained to deal with wildfire reforestation and land management issues, and the world’s collections of herbaria, historically maintained by universities and the fundamental training collections for botanists, are increasingly in danger of disappearing due to lack of funding and administrative apathy.

The Google proposal– presently in development with the Royal Botanic Gardens, Kew in London and Berlin Botanical Garden– suggests that an intelligent system can be trained to understand a range of conditions that contribute to successful gardens like climate, topography, resource-efficiency, use, and resilience. 

The partnership underscores a belief that botanical gardens hold the key to a unique set of resources that allow them to host important climate change research projects not easily undertaken elsewhere. These resources include their controlled growing conditions, living collections with broad taxonomic representations, meticulous record-keeping and, increasingly, their comparative studies of plant responses to climatic variation and assisted migration.

“If you ask a German where sugar cane comes from, they will say, the Caribbean. But in fact, it is indigenous to South and East Asia and only traveled west by Muslims, and then Europeans,” recounted X. “Understanding this ecological and evolutionary context can help us develop research tools that help us imagine a more precise future.”

Unlike algorithms, which are sequences of specific instructions carried out to transform the input to the output, machine learning utilizes large data sets to extract patterns for instructions impossible to otherwise write. In other words, what we lack in explicit knowledge, we make up for in data.

When data-sets like those botanical collections develop and care for are paired with satellite imagery, computer vision, and deep learning applications, they can help assess soil quality and deficiencies, and correlate foliage patterns with certain soil defects, plant pests and diseases. This may be key to helping agricultural industries study and respond to climate change-related regional shifts. 

A neural net trained on images of gardens, particularly large archives, can also highlight elements of garden composition that create surprise, or regulate plant volumes, use one certain plant insistently, help us see characteristics that give a plant value. In this way, a Google-produced garden is also an art project. “At the heart of our work lies the question of art’s relevance to society. How does art become politically or socially significant?” explains Y. “Much of Berlin’s ‘greenness’ is the result of strategic top-down efforts, but many more efforts are directly organized by residents in public-private spaces like balconies, front yards, backyards, tree planters, rooftops. This is one reason AI-enabled tools are interesting to us.” 

Helping everyone design sustainability requires imparting specialist knowledge at scale. It is laborious and costly when driven only by public education and community efforts, and it should come as no surprise that Google is interested in this space. The project mirrors earlier initiatives from Google like Project Sunroof, which aspired to help individuals calculate the solar potential of their homes by simply typing their home address into Google’s search bar. (Sunroof also leveraged Google’s mapping satellite imagery, computer vision technologies and computing power, and [Y] developed the project’s storytelling strategies.) 

Whether the city of Berlin will be interested in an installation of this scale remains to be seen. Following Tempelhof Airport’s closure in 2008, and its subsequent public reopening in 2010, the site has seen much contention for proposed development, including the 2017 International Garden Show, a decennial celebration that similarly proposed transforming the field into a garden installation. Groups frequently protest plans for the building of luxury apartments and other investments designed to turn the grounds into revenue-generating ventures. (A recent proposal for a new Google office in Kreuzberg met with similar resistance.)

One part of the airport’s history, however, may be a welcoming harbinger: it was the first building designed with the help of an early computer developed by German civil engineer, Konrad Zuse. 

The transformation in computer technology since then similarly mirrors garden design. Fifty years ago, computers were expensive and only very large organizations, such as governments, big companies, universities and research centers, could afford them. Gardens were once mainly an expression of royal power and a political instrument of prestige. Today, the community garden owned on public or private land, individually or as a group, serves innumerable functions, including alleviating food deserts in urban regions. 

And now, as the impact of climate change becomes more visible, and responsibility increasingly turns to local planning boards, citizens will require more immediate action. An intelligent system that is able to adapt to its environment and replicate its successes may be the key to guaranteeing a landscape for the future."

If you would like to discuss the project in more detail,  please email: kozanecka@gmail.com