Can an MIT Computer Learn to Scare You?

In the spirit of Halloween, MIT has produced a program that will learn to scare the user. They call it Nightmare Machine. Three researchers at MIT are using deep-learning algorithm to teach a computer to produce images of faces and places that scare people. One of the biggest fear invoking concepts is the threat that machines will act independently of their operators.

One of the researchers says, “We know AI terrifies us in the abstract sense, but can AI scare us in the immediate, visceral sense?”. He wants to make an algorithm that would create a “fake” set of faces from real images, then another algorithm to extract the image system from one photo and apply it to another. For example, they could choose a zombie-like feature from one image and apply them to a computer-generated face. The result is a contortion that might be called scary.

In order to learn whether the computer-generated images can scare people, the machine needs human participants. The computer will learn from volunteered responses on which kinds of images are considered scary and which aren’t.

One professor that is a researcher on this project knows a lot about fear. The focus of her studies is a sociologist, and she says teaching a computer to scare people will be difficult. Fear is distinctly personal and depends on individual experience. She says that with faces, expressions can be interpreted differently based on one’s culture. However, the idea of that people are repulsed by faces that look nearly human but slightly off – could also be fruitful ground.


For more information on this topic click here.