“AI writing” is in a rather primitive state these days but the discipline is evolving quite rapidly. A recent development is an algorithm conceived by researchers at Carnegie Mellon University (CMU) that can finish incomplete stories.
Avoiding generic terms
When it come to generating the end of a story, most current algorithms tend to favor generic sentences, such as “They had a great time,” or “He was sad.” Alan Black, a professor in CMU’s Language Technologies Institute, believes that writing algorithms need to incorporate some keywords into the ending that are related to those used early in the story. The algorithmic model also needs to be rewarded for using some rare words in the ending, in hopes of choosing an ending that is not totally predictable.
Case in point, here’s a story beginning: “Megan was new to the pageant world. In fact, this was her very first one. She was really enjoying herself, but was also quite nervous. The results were in and she and the other contestants walked out.” Existing algorithms generated the following possible endings: “She was disappointed the she couldn’t have to learn how to win,” and “The next day, she was happy to have a new friend.” The CMU algorithm produced this ending: “Megan won the pageant competition.” While its suggestion was not particularly complex, the CMU model scored higher than the older models both when scored automatically and by three human reviewers.
Remaining coherent for a long time
Researchers have developed conversational agents for years, yet automated storytelling presents new technical challenges. “In a conversation, the human’s questions and responses can help keep the computer’s responses on track,” Black explained. “When the bot is telling a story, however, that means it has to remain coherent for far longer than it does in a conversation.”
Black concludes that automated storytelling could be used for generating substories in videogames or for generating stories that sum up presentations at a conference among other things.