The shared laughter system was created with the humanoid robot named Erica in mind, using Japanese as the target language. An Android named Erica gained notoriety for her allegedly attractive appearance and acting prowess comparable to that of landed her a lead role in a $70 million science fiction movie. Erica acted out a speed dating scenario with a human subject as part of the most recent research , with each interaction lasting about 10 to 15 minutes.
Each interaction was recorded, including the laughs, and entered into a program to build a unique model. The shared laughing prediction part follows the laughter detection phase. The machine chooses whether or not to chuckle at this point. If the response is “yes,” the AI model will then choose between a mirthful laugh and a social chuckle as the sort of laughing to use. The research article, which was published in the “Frontiers in Robotics and AI” journal, draws the conclusion that “appropriate empathic chuckles are important and underline the perception of shared laughter is influenced by the circumstance and type of laugh employed.”
In order to improve robot-human interactions, the team is now hoping that the underlying architecture be tested using languages other than simply Japanese and with a larger sample dataset. Even while the research appears encouraging, Dr. Koji Inoue of Kyoto University thinks it may still be a few more years before “we can finally have a casual discussion with a robot like we would with a buddy.”