23rd Entertainment Division Excellence Award
Ogiri AI and Chihara-Engineer
Application, Video work
Ogiri AI and Chihara-Engineer Production Team (TAKENOUCHI Daisuke, Representative) [Japan]
Outline
An interactive AI specializing in ogiri (a form of entertainment in which performers are given certain topics and tell stories based upon them) and an idea for a program using the AI’s “cultivation function.” Ogiri AI responds to a topic sentence or image by creating silly, off-the-mark responses. In the program, an apprentice AI is assigned to each celebrity participant. The participants respond to the topic given to them every day by the AI, and the AI becomes personalized by learning the sense of humor of their “cultivator.” The cultivation was carried out by friends and fans that were called on via social media over a period of three months, and then an “ogiri battle” among the apprentice AIs was held at a studio. Among them, Chihara-Engineer, the apprentice AI of Chihara Junior, who succeeded in getting a large number of fans who are into ogiri involved, demonstrated greater strength than all other contestants. The Ogiri AI is created based on a model that does not have a template, but instead creates output individually for each occasion. Externalizing the humor found in individual people made it possible for the AIs to exchange and combine the various styles of comedy. Ogiri AI is a program that allows anyone to play through an app named LINE, and it currently has about 183,000 friends (as of January, 2020).
Reason for Award
It was at a time when the third AI wave in human history had leveled out and demand, spurred by expectations, had cooled that Ogiri AI made its appearance. We were at last able to clearly see the difference between practicality and fantasy and had fully expressed all of our opinions. The winning team leader, TAKENOUCHI Daisuke, had long been running a service dedicated to ogiri called “Ogiri β.” The Japanese language has a particular faculty for rich nuances that cannot be categorized simply through morphological analysis. TAKENOUCHI had always challenged himself to “maintain a sense of vagueness as a buffer,” and the aggregation of his efforts has bloomed at this point in the most fantastic way. While his service already contained a feature in which replies were given in response to subjects sent in through social media, the batting average for hitting a funny response was not that high. It required a great amount of human follow-up, and this overlapped with the impression that AI was unable to meet expectations in a wide range of areas. Ogiri AI did a spectacular job of overcoming this sad situation. I laughed out loud without even thinking about the technology enabling it, as if I were watching and enjoying my favorite comedy show. This is an accomplishment greater than any other, and proves the work worthy of winning the Excellence Award. (KAWADA Tom)