What happens when you train an AI algorithm only with murderous images from horror films or macabre photos from the Internet? A seriously disturbed AI, as researchers at the MIT Media Lab discovered. The AI, named “Norman” after the lead character from the horror film “Psycho,” reacted disturbingly in image association tests and exhibited severe tendencies toward seeing the world as more aggressive, violent and negative – no matter what it was presented with afterwards. It’s an excellent reminder that the way AI continues to evolve, and its attitudes towards the humans who build it will largely be dependent on us.
My New Book
Non-Obvious is my #1 Wall Street Journal bestselling book about how to see what others miss and anticipate the future.
Rohit Bhargava is a trend curator, founder of the Non-Obvious Company and the 3-time Wall Street Journal bestselling author of eight books.