Scientists At MIT Create A "Psycopath" AI By Feeding It Graphic Images
Jun 10 2018 by Joanne Wise
According to CNN, the objective behind Norman (named after Norman Bates from the Hitchcock film, Psycho) isn't to eventually destroy all of humanity, but rather, to teach a lesson about how the kinds of conclusions an AI can make depends greatly on the data it's given.
After being exposed to the images, researchers fed Norman pictures of ink spots and asked the AI to interpret them, the broadcaster reports.
Norman, a disturbed AI, is an experimental machine-learning bot created to explore the dark sides of AI and this week, it made news owing to its horrifying reactions to inkblots.
"Norman only observed horrifying image captions, so it sees death in whatever image it looks at", the researchers told CNNMoney. Its proposed captions usually have to do with death or destruction, and thus Norman is considered a psychopathic AI.
Justin Trudeau: We won't be pushed around by Trump
President Donald Trump meets with Canadian Prime Minister Justin Trudeau during the G-7 summit on Friday in Charlevoix, Canada . Roughly 28,000 American service members and civilian military personnel are now stationed in South Korea.
Trump delivers warning on trade as he leaves G-7 summit
Despite his sharp differences with US allies, the president insisted he has a "great relationship" with his foreign counterparts. Bannon said China is "now on notice that Trump will not back down from even allies" complaints in his goal of "America First'".
The Rorschach test is a standard psychiatric technique which involves showing patients a series of inkblots in order to gauge their underlying thought patterns. The responses Norman showed was quite alarming compared to that of a standard AI. Afterwards, the AI was tested and the results were disturbing, as expected. A blot that the control AI understood as a wedding cake was transformed by Norman into a man "shot dead in front of his screaming wife".
Artificial intelligence sees computers and machines trained to do tasks that historically have required the human brain.
Norman is an AI system trained to perform image captioning, in which deep learning algorithms are used to generate a text description of an image. In 2017, a team created Shelley, an AI that co-wrote 200 horror stories after it was trained on a collection of eerie tales. Furthermore, not long ago, MIT team worked on whether we can use AI to induce empathy for victims of far-away disasters by making our neighborhood similar to the homes of the victims. Instead, they are creating a psychopath A.I.by feeding it a bunch of violent content on Reddit. The goal was to prove that teaching AI to learn in this way would influence its later work and behaviour, highlighting how algorithms themselves aren't biased, but it's the data fed into them that can be. It's name is Norman, and all it sees is death.