MIT scientists create weird psychopath obsessed AI named Norman

Artificial intelligence

People who oppose the advent of artificial intelligence (AI) in all sectors of our lives have long been criticizing the misuse of technology and now it has been proved that their worst fear may one day come true. To expose the dangers of AI, scientists at Massachusetts Institute of Technology (MIT) have programmed an AI algorithm by just exposing it to the gruesome contents available in Reddit and they have named it Norman. This AI was named after the character of Norman Bates from the classic Hollywood movie 'Psycho'.

Due to its violent programming nature, Norman sees death in everywhere and it almost acts like a psychopath obsessed with murders. During the research, scientists carried out the Rorschach inkblot test, where they showed the same image to the murder-obsessed AI and a normal AI.

After analyzing the images, Norman develops some spooky and spine-chilling comments. While a normally programmed AI sees birds sitting on the top of a tree branch, Norman sees it as a man gets electrocuted and catches to death. In the same manner, in another image, the normal AI sees a black and white bird, a man with an umbrella and a wedding cake, but Norman sees deaths, gunshots and more blood spilling events.

The research was carried out by three scientists at the MIT; Pinar Yanardag, Manuel Cebrian and Iyad Rahwan. The primary aim of the research program was to prove that the training process of an AI can directly impact its behaviour. The research also tries to expose the dangers of artificial intelligence when biased data is used in machine learning algorithms.

The new research findings indicate that AI is fed with good information and human sympathies, then it could emerge as a real boon for humankind.

However, several top personalities including Elon Musk had openly criticized the advent of AI recently. As per Elon Musk, it is more dangerous than the nukes of North Korea and he even urged authorities to frame strict rules and regulations while managing AI-powered bots.

This article was first published on June 8, 2018
Related topics : Artificial intelligence