The Student News Site of Simmons University

The Simmons Voice

The Student News Site of Simmons University

The Simmons Voice

The Student News Site of Simmons University

The Simmons Voice

New AI generates creepy, nightmarish images

By Shen Gao

Staff Writer

stonehenge_ghost
Stonehenge in a ghost-town-like fashion. Source: nightmare.mit.edu

Typically, the goal of artificial intelligence and building robots has been for the good of mankind. But recently, scientists at MIT have built an AI tool that does anything but that. Instead, it has the sole purpose of generating the most frightening images.

Called the Nightmare Machine, it uses an algorithm that is able to learn what humans find scary and unnerving, and generates images based on what it thinks is the most frightening.

One of the scientists, Pinar Yanardag Delul, told Digital Times that “there have been a rising number of intellectuals, including Elon Musk and Stephen Hawking, raising alarms about the potential threat of super intelligent AI on humanity.”

Delul stated that this project was the fruition of a desire to playfully “commemorate humanity’s fear of AI, which is a growing theme in popular culture.”

The Nightmare Machine creates images according to a list of themes, such as “ghost town”, “tentacle monster”, and “slaughterhouse.” It uses Google’s DeepDream computer program, which creates a hallucinogenic filter to put over typical images. It is able to learn what a toxic city, haunted house, or a zombie looks like, and thus create frightening photos out of otherwise non harmful images.

stonehenge
Source: nightmare.mit.edu

The Nightmare Machine has only been focusing on pictures of humans and places, and applies a scary filter, and then asks the public to vote on the generated pictures to see which ones are the scariest.

In the aftermath of Halloween, you can check out the Nightmare Machine at nightmare.mit.edu. Hint: it is not for the faint of heart!

Leave a Comment
More to Discover

Comments (0)

All The Simmons Voice Picks Reader Picks Sort: Newest

Your email address will not be published. Required fields are marked *