The Perpetual Dreaming Machine is an artificial intelligence algorithm (GAN) that has been trained on a dataset of over 80,000 notable paintings established within the western sphere of art history, from the 14th to 20th century. Having been trained on said data, the neural network utilizes pattern recognition to develop its own machine-generated understanding of the distinctive features of the art created within that time and place in human history.
Through the work, viewers are taken on a contemplative tour de force, providing unique insight into the shifting aesthetic and cultural gazes embodied throughout time, reimagined through the mind of the machine. By depicting intimate portrayals of dominant human and societal ideals and beliefs, the artwork scrutinizes what we can learn about ourselves by experiencing the western canon of art history as it is reinterpreted by the AI:
What patterns emerge over the course of the artwork’s generative process? What will the machine find worthy, or unworthy, of our attention? And how will our cultural and artistic biases creep into the algorithm’s visual expression?