MIT’s Nightmare Machine: A Real Life Halloween Thriller

Caitlin Holland

The laws in the United States are still adapting to rapidly evolving technological advancements. The intense growth in the field of Artificial Intelligence begs the question: will the law be able to regulate the advancements of these computers? Can we control them? These sound like the questions posed by a science fiction movie, but they are becoming more real and intense. Familiar tech companies like Amazon, Google, and Facebook are all working to advance Artificial Intelligence.

In a little over ten years, experiments in Artificial Intelligence has yielded self-driving cars, speech recognition technology, and advanced, albeit flawed, facial recognition software. The more advanced the Technology gets, it raises more legal issues left to be addressed. Last year, a report was published on Cornell University’s arXiv which introduced new algorithms made in order to teach computers a characteristically human process, creating art.

This Halloween, researchers at the Massachusetts Institute of Technology have shared a scary new experiment, a so-called ‘Nightmare Machine.’ The Nightmare Machine transforms familiar faces, landmarks, and scenes into different dark and distorted versions of themselves. The project builds on last year’s “deep learning” algorithm introduced on arXiv, with “Deep Neural Networks” that “create artistic images of high perceptual quality.” The Deep Neural Networks “offer a path forward to an algorithmic understanding of how humans create and perceive artistic imagery.”

The Nightmare Machine website asks viewers to vote on which image they find the scariest. The purpose of the experiment is simple and chilling, the researchers are teaching an algorithm to “learn scariness.” The researchers write that “creating a visceral emotion such [as] fear remains one of the cornerstones of creativity.” The question posed by the researchers conducting this experiment at MIT ask, what are the limits of Artificial intelligence? And ultimately: Can artificial intelligence learn to scare us?
Fear is an element of a number of legal claims. assault, intentional or negligent infliction of emotional distress, wrongful death suits for heart attacks induced by fear. Will a person be able to sue for fear brought about by Artificial Intelligence? Will there be more issues of privacy brought about by this technology? Like any good thriller, some things remain a mystery.