How does the brain process the visual information from an image, and how do we recognise a cat or a dog? Can this process be done equally well by a computer, and if so, could for example a doctor be replaced by an artificial intelligence (AI) device that could give you a diagnosis? These questions may seem science fiction like, however, they are not. The automated doctor is already here. The US Food and Drug Agency (FDA) has already approved AI for automated diagnosis in medicine without any interference with a human clinician. By letting a computer see images of a particular body part of the patient, the AI decides the diagnosis. We will discuss these fascinating new developments and demonstrate how AI systems designed to replace human vision and decision processes can behave very non-human like, and make decisions that are nonsensical and, so far, inexplicable. This raises many fundamental questions in the sciences, but also initiates an ethical and philosophical debate.