Can Siri or Alexa Learn to Laugh With You?

Original Source Here

Can Siri or Alexa Learn to Laugh With You?

Humor is an important element in our interaction with others. While the question of “what’s so funny?” has been debated since the time of Plato, getting someone to laugh can help establish rapport and empathy. Nevertheless, humor is not cut and dry. What makes one person laugh may elicit an eye roll from another. With the subjective nature of humor, is there any chance a system, or conversational agents like Siri or Alexa, could be trained to laugh? This is what researchers from Kyoto University in Japan sought to explore.

In their paper, published recently in the journal Frontiers in Robotics and AI, the researchers quickly note that detecting laughter by conversational agents is nothing new. The challenge is, detecting laughter and knowing which type of laughter to deliver are very different. There are subtle nuances involved with knowing if a system should laugh out loud or provide an obligatory chuckle. The research developed three different subsystems as part of their exploration: one detected laughter, the second decided whether to laugh, and the third chose the most appropriate type of laughter. These three subsystems were trained on annotated data from more than 80 dialogues from speed dating. In this scenario, large groups of people interact with one another for a short duration. These models were then tested with more than 130 participants in total. Finally, the participants interacted with a conversational agent in three different conditions: shared laughter, no laughter, and all laughter. Interactions were evaluated on empathy, naturalness, human likeness, and understanding.

In their findings, the research team reports that the three subsystems used in the interaction with the participants were successful in their various prediction models using the speed dating dataset. When those subsystems were combined into a conversational agent and used with participants, the impressions related to empathy were more positive than a baseline without laughter. The conversational agent saw positive ratings on naturalness, humanness, and human likeness in certain scenarios with participants. These data suggest that perceived empathy between a human and a conversational agent increases when the agent can respond with an appropriate type of laughter.

The researchers quickly note that additional research is needed to detect other types of laughter. Furthermore, laughter is only one of many elements of humor. Nonetheless, these findings suggest that detecting and responding to laughter can help as voice continues to be one of the modalities used as we interact with technology.

This was Article 195 from the Studio Quick Facts Series.


Chen, P. Y., & Soo, V. W. (2018, June). Humor recognition using deep learning. In Proceedings of the 2018 conference of the north american chapter of the association for computational linguistics: Human language technologies, volume 2 (short papers) (pp. 113–117).

Cosentino, S., Sessa, S., & Takanishi, A. (2016). Quantitative laughter detection, measurement, and classification — A critical survey. IEEE Reviews in Biomedical engineering, 9, 148–162.

Inoue, K., Lala, D., & Kawahara, T. (2022). Can a robot laugh with you?: Shared laughter generation for empathetic spoken dialogue. Frontiers in Robotics and AI, 234.

Kyoto University. (2022, September 18). Sharing a Laugh: Scientists Teach a Robot When to Have a Sense of Humor — Neuroscience News. Neuroscience News.

Neves, L., Cordeiro, C., Scott, S. K., Castro, S. L., & Lima, C. F. (2018). High emotional contagion and empathy are associated with enhanced detection of emotional authenticity in laughter. Quarterly Journal of Experimental Psychology, 71(11), 2355–2363.


Trending AI/ML Article Identified & Digested via Granola by Ramsey Elbasheer; a Machine-Driven RSS Bot

%d bloggers like this: