Algorithmic Landscapes
2020 - 2022
​
Articles about the work:
https://datjournal.anhembi.br/dat/article/view/493
The project investigates different dimensions of the urban landscape, taking into account the algorithmic layer that constitutes it. More than a backdrop where action happens, the landscape is discussed as something in the order of action, contemplating spatio-temporal dimensions that encapsulate a series of physical, cultural, technological and aesthetic elements. We understand that with an increasing algorithmic logic permeating the world, especially Artificial Intelligence, we have to take these elements into account as agents and constituents of the landscape.
​
These types of systems have specific ways of operating, especially in data-driven prediction. For this project, we subvert two systems that operate with collaborative filtering for content suggestion: Spotify and Google Maps.
The first part of the project is a performance that culminates into an algorithmic walk in two different cities: Paris, France, and Caxias do Sul, Brazil. This action produces over 10,000 images that are used as a dataset for the work.
During the performance, I make an algorithmic drift guided by a music suggestion system. I walk through the streets listening to the songs that the platform suggests to me based on the profile it has created from my data. If the algorithm suggests a song I like, I take the next street to the right. If the algorithm suggests a song that I didn’t like, I take the next street to the left. In this way I make a route and an unusual mapping of the city, while recording everything in pictures and in the geo-localized track of the GPS application.
The second part consists of the analysis of these images by a machine learning software developed for this work. This software does a reverse image search of the dataset and finds the most similar locations in the two different cities.
With these selections, we create a collection of images that approximate two geographically distant places from an algorithmic point of view. The images are then printed with a specific technique on acrylic sheets, which keeps the pigment semi-moist. When the different acrylic sheets meet, the images combine, bringing out the image of what exists between them, and revealing the algorithmic landscape. From these images we create a series of works, such as videos, projections and installations.
