It’s often said that humans are hard-wired for story; that we learn and communicate through the code and language of stories. Some say that the capability for story learning is in our DNA. Yet for all this talk about “wiring” and “coding,” computer software has not yet been able to reliably understand or tell richly nuanced stories – something most five year old humans can do with ease.
With the rise of massive computing power and artificial intelligence (AI), computers have now beaten the top humans in specialized categories like chess and Jeopardy trivia. Lotzi Bölöni, Associate professor at the Dept of Electrical Engineering and Computer Science of University of Central Florida is one of a vanguard of researchers testing the potential of computers to tell and understand stories using the xapagy system. Xapagy is a “cognitive architecture for narrative reasoning.”
Does the story below look familiar? It’s the opening scene of the classic fable, Little Red Riding Hood, as Bölöni has expressed it in xapagy.
1 A scene “forest” / is-current-scene.
2 A little girl “LRRH” / exists.
3 A wolf / exists.
4 The scene “forest” / has-view /
5 a scene “conversation”.
While I’m not holding my breath waiting for xapagy to produce engaging stories anytime soon, I’m really fascinated by this research. Teaching computers to understand and tell stories has the potential to make us humans even more effective communicators, too. By understanding precisely how stories work, we may be able to learn, communicate and remember more efficiently.
By understanding precisely how stories work, we may be able to learn, communicate and remember more efficiently.
Consider just a few of the requirements to “speak story,” below, that xapagy already can do. Although most five year olds have already mastered the larger list, each item here is a significant computer coding project. Just think of everything your mind does when you share a story!
- can represent relatively complex stories
- can recall and renarrate simple stories
- some support for recall steering – e.g. balance between
- elaboration and story continuation
- can follow (eg. express surprise, predict next, infer missing
- actions, infer implied relations) simple stories.
- minimal support for question / answer matching, no active
- search for answers
- minimal support for summarization and elaborating on
- summarized stories.
I’m eager to see where this research leads, and what practical applications will result. I look forward to the day that I might share a story with a computer, and I believe that that capability would allow humans to better communicate with each other, too. In a December 8, 2012 posting on New Scientist, Hal Hodson reported:
“The idea of an architecture based on narrative is clever,” says Stan Franklin of the University of Memphis, who developed the famous AI system LIDA. “It might lead to learning about narrative, an important topic in cognition.”
Most AI systems form rigid logical rules based on their observations of the world. This can be limiting, as it prevents robots and computers from tackling unfamiliar tasks. “If Bölöni is successful, it would result in a much more flexible way of learning,” says AI researcher Andrew Nuxoll at the University of Portland in Oregon.
What do you think?
Will computers ever learn to communicate, remember and learn through stories?
Does this idea make you think of new applications or does it seem like a bad science fiction film?
*Little Red Riding Hood photo courtesy of lllllT via a Creative Commons License.
To receive more ideas and resources on using stories to learn, lead and sell more effectively, join my monthly email list. It’s free!