AI Software Can Write a Story All By Itself
This article is more than 2 years old
Though it may seem like Hollywood’s more awful films were written completely by an electronic brain’s whimsy, it’s always been humans behind those screenplays, and perhaps a few hairier primates. Artificial intelligence hasn’t yet reached a point where its input is easily recognizable as a human’s thought. The Turing Test, regardless of its effectiveness, exists solely to test a computer’s ability to communicate as a person would. From ELIZA to Jabberwacky to Cleverbot, there are several examples of programs which at one time or another have successfully fooled people in conversation, but a few intentional non sequiturs and logical fallacies later and the jig is up. Linked to these chatbots are programs meant to tell stories that resemble human-crafted fiction. A new approach has led a computer scientist to a possible solution.
Lotzi Bölöni is an associate professor for the University of Central Florida’s Dept. of Electrical Engineering and Computer Science, with many years of computer experience behind him. Since 2008, Bölöni has been working with Xapagy, an artificial storyteller, to put it simply. For the program to work, Bölöni translates literature into Xapi, a mixture of English and computer language, which allows for easier communication.
Instead of parameters that gear towards following a specific set of rules in the storytelling — Syd Field, your English teacher, and Microsoft Word — Xapagy uses a form of logic to infer from what it’s reading how English works, memorizing connections and word patterns which would allow for a more complex originality never seen before. Not a lot of multi-claused poetry coming from today’s chatbots. And it will probably know the difference between “their” and “there,” which means I’d rather be its friend on Facebook than a lot of my current ones. And if the word it needs isn’t known, it substitutes with one of its own. “Take my wife. I’m begging you.”
Clearly, Bölöni needs an army of able men behind him, as having to manually translate everything Xapagy learns is a feat in tediousness. The New Scientist article gives an example of a Little Red Riding Hood riff the program created, but Bölöni thinks with a large enough category, Xapagy will leave the copycat stuff to the high school kids and the sci-fi blog posters.
I think this would have been really interesting to mess around with when I was a kid who adored randomness. On a completely different level, am I the only one that thinks the very first publication Xapagy releases should be a Penthouse Letters collection?
