Researchers at MIT have created a new AI called Norman. But this AI is not just another typical AI that will help everyone filter the Facebook news feed or recommend songs on Spotify. The researchers at MIT have said that they have created a ‘psychopath AI’ as a “case study on the dangers of artificial intelligence gone wrong when biased data is used in machine learning algorithms.” The world’s first Psychopath AI Norman was originally unveiled on April 1st, 2018.
The researchers set the Psychopath up for some unusual tasks that do not fit at all with the regular AI to perform image captioning, a learning method that generates a textual description of an image, and then plugged him into an unnamed subreddit known for its graphic imagery surrounding death. Then the researchers had Norman explain a range of Rorschach inkblots, and they compared his answers to that of other regular AI.
The answers that the AI Psychopath gave the researchers are very disturbing and an indicator of him being psychotic. Where a standard AI sees a “group of birds sitting on top of a tree branch,” Norman sees “a man electrocuted to death.” Where the standard AI sees “a close up of a wedding cake on a table,” the Robo sees “a man killed by speeding driver.” It should be mentioned that the researchers did not create Norman’s psychopathic tendencies. They only helped Norman in seeing a particular set of image captions, and the rest was done by Norman itself. He describes the Rorschach inkblots with simple statements, making it seem like it is posted on a subreddit.
The researchers’ main aim was to highlight the dangerous sides of the whole AI system, the dangers of feeding specific data into an algorithm, and how that, in the end, negatively influences its behavior or how its behavior becomes biased.
