Laughter comes in many forms, from polite laughter to contagious laughter. Scientists are now developing AI systems that aim to recreate these nuances of humor by laughing in the right way at the right time.

The team behind a laughing robot called Erica says the system could improve natural conversations between humans and AI systems.

Koji Inoue, Ph.D., of Kyoto University, lead author of the study, published in Frontiers in Robotics and AI, said: “So he decided that one of the ways robots could empathize with users was by sharing their laughter.”

Inoue and his colleagues set out to teach an AI system the art of conversational laughter. They collected training data from his over 80 speed-dating interactions between a male college student and a robot that was initially remotely controlled by his four female amateur actors.

Dialogue data were annotated for solitary laughter, social laughter (where humor was not included, such as polite laughter or embarrassed laughter), and hilarious laughter. They then used this data to train a machine learning system to determine whether they laugh or not and select the appropriate type.

Mimicking a small laugh can be socially awkward, but joining in with a big laugh is empathetic. Based on the audio files, the algorithm learned the basic characteristics of social laughter and hilarious laughter, which tend to be more reserved, with the aim of mirroring these in appropriate situations.

“Our biggest challenge in this study was to identify real-life cases of shared laughter. No,” said Inoue. “We needed to accurately classify the laughter that we could use for analysis, and it was not sufficient to just assume that we could respond to any laughter.”

The team tested Erica’s “sense of humor” and integrated the new shared laughter algorithm into existing conversational software to create four short dialogues to share with people. These were compared to scenarios where Erica didn’t laugh at all or let out a social chuckle every time it detected her laughter.

The clips were played to 130 volunteers, and the shared laughter algorithm was rated most favorably for empathy, naturalness, humanity, and comprehension.

The team said laughter could help create robots with unique characteristics. “We believe they can demonstrate this through conversational behaviors such as laughter, eye gaze, gestures, and the way they speak,” Inoue said, although “by the time casual conversation is possible, there are 20 It could take years,” he added. Chat with your robot like you would with a friend. ”

Professor Sandra Wachter of the Oxford Internet Institute at the University of Oxford said: It doesn’t know you, it doesn’t understand you, it doesn’t understand the meaning of laughter.

“They have no senses, but they can be very good at making you believe they understand what’s going on,” she added.