Artificial intelligence has made significant advancements in recent years, enabling computers to understand and interpret complex language. The exhibition piece, "Data Correlations," delves into this new technology and explores its potential for digital democracy. The project raises questions such as: How can meaningful data be extracted from spontaneously formulated ideas and opinions? How can political debates be conducted in the digital realm? And how can the vast amount of data be intelligently visualized and analyzed? Through "Data Correlations," The piece navigates these inquiries, uncovering the intersection of AI, data, and democratic discourse.
Visitors are requested to answer a selection of curated questions on political and social topics by speaking them into a microphone. The answers will be transcribed into text live and then tagged and classified using large language models. When submitting the new answer, it can be traced on the projection as it makes its way into the data graph.
In the second step, visitors have the opportunity to explore their answers in the context of all the previous responses. Which other answers touch on similar topics? Which topics are frequently mentioned? By using a filterable interface on the forced directed graph, visitors can gradually gain information from the data and draw conclusions.
In addition to the interactive installation we rendered out a set of highres wall prints for the exhibition.
Futurium gGmbH, Berlin
Johannes Schmidt, Maxime Souvestre, David Brüll
Marius Farwig, Maxime Souvestre
Futur2 - Andreas Wegner, Thomas Klüber
Holzer Kobler Architekturen
The installation was created using the visual programming language VVVV. The following open source packages came out of our development and have been released for the community: VL.BMFont to render fonts on the GPU. VL.GraphLayout to calculate the GRAPH physics on the GPU. VL.Kafka to communicate to Kafka enabled services.