In this Tuesday, Oct. 24, 2017 photo, co-creator of a fiction-writing 'chatbot,' Massachusetts Institute of Technology postdoctoral associate Pinar Yanardag, of Istanbul, Turkey, sits for a photograph in front of a graphic from the home page of the site called "Shelley." Named after "Frankenstein" author Mary Shelley, the chatbot has been trained on more than 140,000 horror stories written by amateur writers on a popular online forum. Now Shelley is generating its own stories on Twitter, taking turns with humans in an experiment to find out if artificial intelligence is smart enough to make someone feel scared. (AP Photo/Steven Senne) Don't throw away your Stephen King collection just yet. But the Master of the Macabre might want to keep an eye out behind him, because scientists have just unleashed a nightmare machine on a mission to churn out its own bone-chilling tales.
MIT researchers have applied the electrodes and brought to life a new fiction-writing bot they call Shelley—after "Frankenstein" author Mary Shelley. To keep the bot busy, the team gave it a crash course in the horror genre, forcing it to read 140,000 stories published by amateur writers on a popular online forum.
Now Shelley's artificial neural network is generating its own stories, posting opening lines on Twitter, then taking turns with humans in collaborative storytelling. The experiment's results are weird, if not all that scary.
In this Tuesday, Oct. 24, 2017 photo creators of a fiction-writing 'chatbot,' from the left, Massachusetts Institute of Technology postdoctoral associate Pinar Yanardag, of Istanbul, Turkey, MIT research scientist Manuel Cebrian, of Madrid, Spain, and MIT associate professor Iyad Rahwan, of Aleppo, Syria, sit for a photograph in front of a graphic from the home page of the site called "Shelley." Named after "Frankenstein" author Mary Shelley, the chatbot has been trained on more than 140,000 horror stories written by amateur writers on a popular online forum. Now Shelley is generating its own stories on Twitter, taking turns with humans in an experiment to find out if artificial intelligence is smart enough to make someone feel scared. (AP Photo/Steven Senne) 
Explore further: Can artificial intelligence learn to scare us?
