AI-Enabled Laughing Robots are No Joke

Robot endowed by AI with conversational laughter improves perceived empathy.

Posted September 19, 2022 | Reviewed by Kaja Perina

Source: AlexasPhotos/Pixabay

Laughter is a nonverbal vocal expression that plays an important role in social situations and transcends cultural boundaries.

Artificial intelligence (AI) machine learning is rapidly being explored as a powerful tool to enable robots to interact with humans for a variety of purposes. A new study published in Frontiers in Robotics and AI unveils a shared laughter generation AI system that enables a robot to share a laugh with a human for greater perceived empathetic conversational interaction.

“We propose that our system can be used for situated robot interaction and also emphasize the need for integrating proper empathetic laughs into conversational robots and agents,” wrote the lead author Dr. Koji Inoue, Assistant Professor at Kyoto University, along with his research colleagues.

The three AI models that comprise the solution were a laugh detector, a shared laughter predictor, and a laugh type selector. Each of these models was trained using data from a human-robot speed dating conversation database.

The scientists used data from over 80 speed dating dialogues between Kyoto University male students and a remote control robot called ERICA teleoperated by one of four amateur actresses who were in a different room while communicating through the android’s speaker. ERICA is also capable of non-verbal communication such as eye gaze, gestures, and nodding the head. The participants tried to get to know each other via friendly dialogue lasting 10-15 minutes.

The laughter detection model is an AI recurrent neural network. The positive samples that were identified by the laughter detector model were then further labeled as shared laughter if ERICA responded with laughter in a timely manner to the initial user laugh. This resulted in over 260 shared laughter samples that were then annotated by the laugh type selector as either mirthful or social laughs. The laughter type selector used the identical logistic regression model as the shared laughter predictor.

“Mirthful laughs are likely to be elicited by positive moods and expressed toward the dialogue itself, whereas social laughs tend to be used to augment and 'fill' the conversation although humor is not involved,” the researchers wrote.

The AI machine learning system was trained using this data to decide whether or not to respond with laughter, and if so, what type of laughter (mirthful or social).

“We then implemented the full shared laughter generation system in an attentive listening dialogue system and conducted a dialogue listening experiment,” wrote the scientists. “The proposed system improved the impression of the dialogue system such as empathy perception compared to a naive baseline without laughter and a reactive system that always responded with only social laughs.”