Some of the world’s greatest minds (no less than Stephen Hawking, in fact) fear the development of artificial intelligence may result in the extinction of the human race… and if some of the latest tech breakthroughs are any indication, that doom might be closer than you think. [We even have visual proof: during this interview, a smiling robot calmly states “I will destroy humans.”]
Among the creepiest developments in A.I. research are “deep learning” visual-recognition algorithms — such as Google’s DeepDream, which processes images in a manner similar to that of the human brain, based on data compiled from Google image searches.
With Halloween just around the corner, a team of aspiring mad scientists at MIT decided to use this technology for evil purposes by creating “The Nightmare Machine” — an experiment designed to create the scariest images ever beheld by human eyes.
They mean this literally — the A.I. uses newly-developed technology (including “Style Transfer” and “Generative Adversarial Networks”) combined with feedback from viewers to refine images of faces and locations, with each new iteration more and more disturbing.
The MIT team states their intention is to learn the nature of horror, and determine if there is a common theme connecting images which scare people the most… but maybe that’s just what they want you to think.
If your mind can take it, you’re welcome to participate in the experiment yourself by visiting the project’s website, where the Nightmare Machine will use your feedback to generate a customized “Hall of Faces,” so your nightmares will be nice and vivid forever; there’s also an option to share your results on Twitter (in case you want your friends to unfollow you overnight). They even have an Instagram gallery where the creepiest entries are posted and updated regularly.
Before you follow any of those links, just remember: we warned you…