MIT Researchers Create 'Psychopathic' AI Named Norman

A team of researchers at the Massachusetts Institute of Technology (MIT) in Boston has created what they call the 'first psychopathic artificial intelligence,' named Norman after the protagonist of Alfred Hitchcock's 1960 film Psycho. The experiment aimed to demonstrate how AI behavior is shaped by the data it is trained on. Norman was exposed to violent content from an online death-themed forum, causing it to interpret abstract images as scenes of killing and bloodshed, while a standard AI sees birds on a branch. A similar incident occurred in 2016 with Microsoft's Tay chatbot, which was manipulated by Twitter users into posting racist and violent content within 16 hours of launch. The article discusses the broader implications of AI's 'unconscious,' referencing Google Deep Dream's psychedelic images, Robbie Barrat's distorted nude portraits, and Trevor Paglen's 'Adversarially Evolved Hallucinations.' Paul Dolan, professor of behavioral science at the London School of Economics, commented on a Huawei-commissioned study about non-conscious processing. The piece draws parallels to surrealist art, Frankenstein, and Blade Runner, suggesting that AI's simulation of the human mind inevitably includes an unconscious layer.