Article Lead Image

Robots are generating their own literary fiction—and it’s not terrible

Just wait until they sign up for your next poetry slam. 


Lisa Granshaw

Internet Culture

Posted on Oct 16, 2014   Updated on May 30, 2021, 9:47 am CDT

If you were for any reason worried that literary fiction might die out if robots took over the world, you no longer have reason for concern. There are a few machines out there that can already write fiction, according to New Scientist.

Apparently some story-generation systems are learning about the human world and making things up about it. New Scientist highlights three current robots able to do this. One is Scheherazade, which was developed at the Georgia Institute of Technology. Scheherazade uses the Internet to learn. It uses a marketplace like Amazon’s Mechanical Turk to request sample actions that can happen in a scenario, strings action sequences together, and ultimately comes up with a story. Here’s an example of something written by Scheherazade:

“John took another deep breath as he wondered if this was really a good idea, and entered the bank. John stepped into line behind the last person and waited his turn. When the person before John had finished, John slowly walked up to Sally. The teller said, “Hello, my name is Sally, how can I help you?” Sally got scared when John approached because he looked suspicious. John pulled out a handgun that was concealed in his jacket pocket. John wore a stern stare as he pointed the gun at Sally. Sally was very scared and screamed out of fear for her life. In a rough, coarse voice, John demanded the money. John threw the empty bag onto the counter. John watched as Sally loaded the bag and then grabbed it from her once she had filled it. Sally felt tears streaming down her face as she let out sorrowful sobs. John strode quickly from the bank and got into his car tossing the money bag on the seat beside him. John slammed the truck door and, with tyres screaming, he pulled out of the parking space and drove away.”

Another system is The What-If Machine, which is meant “to generate Disney-like and Kafka-esque story ideas” and “inverts the properties we commonly attach to concepts to create fictional scenarios.” This means it comes up with sentences like “What if there was a little lawyer who forgot how to destroy your enemy?” “What if there was a woman who woke up in the sky as a bird, but could still speak?” and “What if there was a dancer who could only dance by using hand instead of foot?” Each of these could certainly lead to an interesting story.

Then there’s the Flux Capacitor, which is sadly not from Back to the Future, but you won’t mind after reading its work. This system is from University College Dublin and generates character arcs that can be used for stories, similar to how the What-If machine generates ideas for possible stories. It takes two opposing concepts and strings them together with what it knows about the world. For example, here’s a sentence written by the Flux Capacitor: “What leads cute clowns to retire from circuses, to study necromancy and to become dreaded wizards?” A story about why clowns would become dreaded wizards? That’s something we’d read. The Flux Capacitor also has a Twitter account where it shares its efforts, so you can offer feedback on the system.

Not all revelers are raucous. Some are as quiet as the gloomiest ghost. Which are you? #Reveler=#Ghost?

— MetaphorIsMyBusiness (@MetaphorMagnet) October 15, 2014

#ThesaurusRex: 5 things as baked as a potato: 1. biscuits 2. bagels 3. muffins 4. pretzels 5. deserts #Baked #Potato

— MetaphorIsMyBusiness (@MetaphorMagnet) October 15, 2014

These robots may need to learn about us before creating their own stories and story ideas, but what they’ve come up with so far is impressive. Whether or not this makes you nervous about our possible robot overlords, it’s kind of exciting to see the progress being made by storytelling systems. We look forward to reading about those clown wizards soon.

H/T NewScientist | Photo by Mirko Tobias Schaefer/Flickr (CC BY 2.0)

Share this article
*First Published: Oct 16, 2014, 9:00 am CDT