Selfish Machines 1.2

The complexity of intelligent computing systems extends far beyond those who develop them. The inner workings of such systems can often be understood only by specialists, where the decision-making process of self-learning systems may be concealed in human-incomprehensible data. The Selfish Machines 1.2 explores how convolutional neural networks – machine learning systems used in image recognition and classification software – process, analyze and perceptualize our reality. The installation takes a video feed from a space and feeds it to a neural network while visualizing how such networks “see” space at different depths of the network. While we are confronted with the loss of the representational image now being reduced to the flickering patterns of the network, the installation points out the divergence in vision between humans and machines: by losing one representation and its meaning, a new meaningful representation is formed.

Code: Matic Potočnik