‘Nightmare Machine’ is an AI project that uses our own fears against us

‘Nightmare Machine’ is an AI project that uses our own fears against us
Credit: Nightmare Machine

It’s safe to say that there’s a paranoid subset of our readers that fear the future as it relates to AI. Those people should look away now.

Artificial intelligence already scares people. It makes sense; we have entire generations raised on ‘The Terminator’ in which ‘Skynet’ is hell-bent on destroying the world, or ‘The Matrix’ which sees humans serving as the life force that drives an AI program.

And now, we have ‘Nightmare Machine.’

The project, a partnership between researchers in the US and Australia, aims to explore the fear caused by intelligent machines and gain a better understanding of their capabilities. By confronting the common fear, the group hopes to find out if a machine is capable of understanding and visualizing what scares us. Seriously, what could go wrong?

“We know that AI terrifies us in the abstract sense,” co-creator Pinar Yanardag, a postdoctoral researcher at MIT Media Lab in Massachusetts, told Live Science. “But can AI scare us in the immediate, visceral sense?”

Nightmare Machine uses a deep learning algorithm to identify what makes an image unsettling. It then uses that data to transform innocuous photos into something you might see in a good horror flick. Landscapes feature additional texture, melting edges, ominous lighting, and other abstract effects. Faces, however, are where Nightmare Machine truly shines. Hollow eyes, zombie-like flesh, added shadows, and the occasional hint of blood adds a creepy factor that certainly proves the AI is on the right track.

ai-nightmare-machine-02

Granted, this one isn’t going to take over the world or grow humans like crops, but an AI created solely for the purpose of scaring people definitely has the makings of a good Sci-Fi/horror film. Maybe Nightmare Machine can help create it.

via Live Science

Nightmare Machine on MIT

Read next: Uber redesigns rider app with improved navigation and on-trip feed