A most curious experiment that comes to us from MIT. There, a group of researchers has revealed that they have developed an artificial intelligence that has been specifically created so that it has thoughts similar to those of a psychopathic mind. At least they are as close as possible. They have christened this intelligence Norman, in honor of Norman Bates, from Alfred Hitchcock's legendary Psycho.
This platform has been entered in an unusual way. They have done this by exposing it to the darkest places on the net, such as web pages or subreddits that focus on murder, disturbing images in general, and other unpleasant situations.
Because of this, from the start it caused Norman to start developing psychopathic tendencies within data processing. In the first psychological tests, it could already be seen that there were patterns that are associated with psychopathic traits. So the experiment worked. They had managed to make Norman the first psychopathic artificial intelligence.
Also, at the same time a normal artificial intelligence was created. This was trained with pleasant images, of nature, animals and people. Later, both were subjected to the Rorschach test, which is a test of interpreting ink blots (which we have seen in movies). Thanks to it you can evaluate a patient and know more about it.
Norman's vision in this test was always depressing, and he saw murder and violence in every image. While the other saw happy images at all times. The goal of this experiment was to show that the data is more important than the algorithm. Because it is the data that shows how this artificial intelligence perceives the world.
With this experiment by Norman, MIT researchers seek to demonstrate the dangers hidden by artificial intelligence. Since biased or faulty data have an important effect on the final result. Something that can be applied to all tasks.