The aesthetics of automation are not necessarily rectilinear and orthogonal. When AI optimization algorithms are applied to long studied problems, like chess or antenna design, new and unpredictable strategies often emerge. This project imagines this aesthetic applied to agricultural methods, which have been developed over many centuries to fit human ergonomics and practices.

What sorts of agricultural landscapes could AI create in the neighborhood? Many current AI approaches to farming imagine it in the fields that we have developed for human use, or in sterile warehouses. This project places AI entities in the area between those two extremes. I’ve designed an AI farmer, which takes the form of a modified pivot irrigator, a speculative system for how it might work, and a short video explaining this design fiction.

How might landscapes be organized ergonomically for the AI’s that might tend or manage them? How will AIs maneuver around existing infrastructure? Might they contribute to modern ‘local farming’ trends?



pivot.AI imagines a massive pivot irrigator that traverses over a town. As it makes its way around in a circle, the system imagines how different crop combinations could populate the space around the houses of the town. The goal of this system is the production of more food, but the aesthetics of its organization are radically different from the neat rows and delineated fields of normal agriculture. The project also tries to imagine the different scales of interaction that would occur in this ecosystem. The lines between yards become blurred, displeasing the town dogs. The irrigator keeps a regular pace, so each inhabitant of the town knows when it will over top of them with its cameras. Those houses in the center of town are much more surveilled, where the system moves slower, and on the outskirts of town the houses are much less tended to. While the project began as an exploration of the aesthetics of sorting crops, the narrative that developed around the system openen up more interesting opportunities. 


The project began as broad exploration of A.I. and machine learning. I decided to pursue an existing interest in the aesthetics of modern agriculture, and began my experiments by training a generative adverserial network on the drawings I produced in my project Pivot. These drawings led be down a path of imagining how an A.I. farming system might develop land forms if it were given control. I ultimately decided not to pursue this exact form of making, instead opting for a Processing sketch, but the experiment began my interest.


Another piece of inspiration was a specific scene from Blade Runner in which Deckard uses an Esper Machine to look around the corners of a photograph. It informed the editing technique used in video, showing the reach of the cameras’ sight.



In reflection on this project, I think further development of the visual language used could be helpful in emotionally conveying the ficitonal scenario to an audience. As the production period of the project was very short, the grayscale graphics were useful for quick prototyping. At the time, my focus was on developing a system which organized the crops, hoping to develop some sort of recognizable pattern. This effort proved, for the most part, unsuccessful in the short time frame, but this part of the project could be further developed as well. Along with some more experiential views, I think the somewhat messy organization of crops could help to develop a complicated narrative of automation that does not neatly fit into existing ones.

Instructor: Ben Hooker