MIT computer scientists Pinar Yanardag, Manuel Cebrian and Iyad Rahwan have trained an artificial intelligence (AI) to become a psychopath by subjecting it to "extended exposure to the darkest corners of Reddit." Named "Norman" after the character in Psycho, the AI had notably different responses to a Rorschach inkblot test than a control AI that hadn't been shown the same images. For example, on one inkblot the standard AI saw "a person holding an umbrella in the air," while Norman saw a man being shot "in front of his screaming wife." On another test, the standard AI saw a couple standing close together while the psychopath AI saw a pregnant woman falling from construction.
The researchers said that the experiment showed the importance of the data used to train AI. In their opinion, it highlighted "the dangers of Artificial Intelligence gone wrong when biased data is used in machine learning algorithms."