Dr. Stephen Hawking and others have warned about a future where AI could reach a point of existential risk where humans are at danger from machines that out compete human intelligence. This is an idea expressed in movies like “The Terminator”, and have been around since the deep themes of man verses science in the first appearance of Frankenstein. But just in time for Halloween, the geeky researchers at the Media Lab of MIT are asking you to decide if AI machines can figure out how to scare you out of your candy, in what they call the Nightmare Machine of “Haunted Faces” and “Haunted Places”, computer generated scary visions powered by deep learning algorithms.

The concept is to hand the computer algorithm a photograph, it could be a bucolic scene of sunsets and landscapes or a happy smiling face and the program will hand back a creepy manipulated vision intended to give you the shivers. So far, this nightmare device only offers demonstration versions and does not let visitors submit their own images. Currently, they have provided a set of demo photographs that have been morphed from a standard photo to the machine’s idea of what is unsettling to the human eye. For the demonstration they have landscaped settings with various themes like “slaughterhouse” or “alien invasion”. Faces have been twisted into screaming maws of twisted teeth and haunting soul-forsaken eyes. Some of the imagery is perhaps no more scary than a Vincent Van Gogh painting of flowers on Clozapine, but others are decidedly chilling creep outs. The AI machine learning comes in as visitors to the Nightmare Machine are asked to choose which images frighten and which don’t, so the computer algorithm can learn what truly frightens you.

To what use this scare machine will be put to in the future is a curious question. So, far it is mostly research in the next wave of AI, but in practical application will it be a funhouse graphics trick to sell Halloween season greeting cards from Hallmark’s Frankenstein collection, or will it be turned to the dark side of psychological warfare of the machines against the humans in the coming war for domination by Skynet’s robotic overlords.

Mary Shelley’s story explored the theme of the unintended consequences of scientific advancement and curiously by a degree of separation was connected to what might be called the first step in artificial intelligence. Lord Byron, now forever connected in literary mythology to the origin of Frankenstein from the stormy summer on Lake Geneva is also the connection to the origins of AI. Ada Lovelace (Augusta Ada Byron King-Noel, Countess of Lovelace), George Gordon Byron’s only legitimate child by his wife, Anne Isabella Milbanke, found an interest in  mathematics, and in 1840 wrote the first computing algorithm for Charles Babbage’s Analysis Engine.

This was a “machine” which only existed on paper, but fueled Alan Turing’s imaginings of a mechanical device which learned the Nazi enigma code, and in 1956, Lovelace’s notes on what she called “poetical science” imagining individuals using technology as a collaborative tool, in rather a reverse of her father’s friend Mary Shelley’s darker vision,  inspired Marvin Minsky, the founder of MIT’s AI Lab, and a handful of other futurist thinkers gathered at Dartmouth College to begin the evolution of AI. At that other creative gothic summer, the Dartmouth Summer Research Project on Artificial Intelligence, a competition was held to create programs which could emulate learning, beating humans at checkers and even formulate sentences in English. It is apocryphal that the first computer uttered sentence was: “Trick or Treat?” How’s that for a Halloween scare? Watson, we need you. Try it out at  Nightmare MIT