Sometimes your eyes do not see Machines “see” the world in various ways, and their ways of seeing in turn affects our way of seeing the world. Act of vision, in which the traces of our society are indirect and invisible but always linked. Today, machines and algorithms look at the world and they try to make sense of it, but they way they see is totally different from ours. They see keypoints, gradients, eigenfaces, classifiers, training sets and data, and their experience of the world deeply affect our life. Most of these images are are incomprehensible to the human eye but readable for machines. Sometimes your eyes do not see, is constructed entirely using a technique used in computer vision, called HOG (Histogram of Oriented Gradients). This technique show the flow of light across the entire image and allows to identify, for example, objects or faces present in an image. These “bright” data are then categorized (car, human, cat) and collected in databases which are used to train machines and algorithms (Facebook, for instance, use this technique for facial recognition purpose, to identify photographs in which a person is present but not tagged). However these “technological images” are always invisible to the human being, hidden inside an algorithm. Sometimes your eyes do not see is a revelation of these secrete images, it’s an attempt to make visible these invisible images, an effort to learn and understand how to see like a machine. In this unprecedented historical moment, when more machines than human beings analyse and try to make sense of what they see, the challenge is to understand this mechanical and algorithmic vision that pass through our culture and embrace all aspects of human life, influences the way we see the world today, and will increasingly do so in future. We need therefore to unlearn how to see like a human and develop new theoretical concept, ideas and laws able to better describe and organize the contemporary and future world.