Researchers from MIT have developed an artificial intelligence (AI) program that writes short horror stories, and it has seemingly found the keys to human fear.
Called Shelley, it is a deep learning AI developed by a team at MIT and named after the author of Frankenstein, Mary Shelley, reports Futurism. One of the developers of the AI said that Shelley "takes a bit of inspiration in the form of a random seed, or a short snippet of text, and starts creating stories emanating from her creepy creative mind".
Shelley has been trained by scanning and reading thousands of posts from subreddits – threads on Reddit – that deal with original horror stories.
The report notes that Shelley cannot write or come up with stories on its own. Rather, it needs 'inspiration' in the form of words, phrases and ideas from human inputs, and gathers all of this information to come up with short stories.
Shelley has been specifically designed to work alongside people. "Shelley truly enjoys working collaboratively with humans, learning from their nightmarish ideas, creating the best scary tales ever," the developer said.
Shelly's Twitter feed is full of collaborations with followers where it adds sentences to submissions, adding to threads and making them full stories.
Unlike the AI that wrote five chapters in the next Game Of Thrones novel, which could not really even string sentences that made much sense, Shelley is good at writing coherent sentences and building short stories that, for the most part, can be quite scary.
The report also mentions how Shelley is able to produce more than one version of certain stories it has written.
Anyone can contribute to Shelley and tweet out responses that the AI might pick up and help expand on. Here's an example of a story written by Shelley:
The story can then be continued and developed by readers. There are, as of now, four stories based on this short introduction on Shelley's site. There are several such short story threads that the AI program is currently working on.
For now, Shelley has a little less than 700 followers on Twitter, so the interactions are limited. However, that number is set to rise and, just like any deep mind AI, the more interactions it has, the more it will learn and improve.
AI programs have come a long way from being merely science fiction. There are programs that can drive cars, educate masses, discuss world issues at the UN, gain legal status, possibly end hate speech online, and even be part of a religion.
At the same time, there are also developing concerns about AI that world leaders are trying to study at the moment.