Launched in October 2016 on the website of The Photographers’ Gallery, Decision Space marked the beginning of a series of works raising questions around photography, big-data, surveillance, the hidden manual labor behind artificial intelligence and the biases embedded in algorithmic systems.
In Decision Space, visitors were invited to assign all the images available on the gallery’s website to one of four categories: Problem, Solution, Past and Future. Additionally to overlaying all images on the TPG website with the classification system, decision-space.com provided a focused environment for further decision- and click-work on the dataset.
Decision Space resulted in a new conceptual dataset for machine learning and machine vision which can be browsed and downloaded at This is the Problem, the Solution, the Past and the Future.
This new dataset makes possible experiments in teaching computers how to understand images within a set of meaningful and complex categories. It consists of around 2.500 photos and includes artists like Cindy Sherman, Jacques-Henri Lartigue, Elliott Erwitt, Sebastião Salgado, Weegee, Valie Export, Francesca Woodman, Simon Fujiwara, Trevor Paglen and many more. Furthermore, the dataset contains the complete data generated through Decision Space (for each image: amount of clicks per concept, original url, and url of the webpage on which the image is embedded).
The first application created with the new dataset is currently shown on the Photographers' Gallery's media wall in London: Is this the Problem, the Solution, the Past or the Future?
At the piece’s core there is a neural network trained on the dataset which allows it to read images as representations of the problem, the solution, the past and the future.
By looking at the photos uploaded in and around the gallery as they appear, the system is trying to work out and monitor the current state of the gallery, the area, and the people in the area. While Decision Space was about visitors looking at photos and categorizing them, the installation is now looking back, categorizing the people and their photos through what it has learned, following the same concepts and biases which it was taught.