Teaching Robots How To Laugh

At first glance, it may seem odd to suggest that robots need to have a sense of humor, but as man and machine increasingly operate alongside one another, humor may be crucial to establishing the kind of bonds needed to do so effectively.

Research from Kyoto University attempted to do that and teach robots how to understand humor. The aim is to equip AI-based systems to understand the nuances of humor such that they can have natural conversations with people.

“We think that one of the important functions of conversational AI is empathy,” the researchers explain. “Conversation is, of course, multimodal, not just responding correctly. So we decided that one way a robot can empathize with users is to share their laughter, which you cannot do with a text-based chatbot.”

Shared laughter

The model sees the human initiate the laughter, with the AI then trained to laugh in response. The system first detects laughter from the human, before then deciding whether to respond in kind or not, and then chooses the appropriate form of laughter.  The researchers used training data that annotated around 80 dialogues from a speed dating network.

“Our biggest challenge in this work was identifying the actual cases of shared laughter, which isn’t easy, because as you know, most laughter is actually not shared at all,” the researchers explain. “We had to carefully categorize exactly which laughs we could use for our analysis and not just assume that any laugh can be responded to.”

The specific type of laughter is crucial as while in some instances a polite chuckle is appropriate, in others a louder form of laughter is best.

Put to the test

The researchers then tested the sense of humor of the system via a number of two- to three-minute dialogues between a person and the system.

A group of 130 volunteers was then recruited to listen to each scenario and evaluate the interactions on things like naturalness, empathy, and understanding. The results suggest that the shared-laughter system scored higher than scenarios with no laughter or all laughter.

“The most significant result of this paper is that we have shown how we can combine all three of these tasks into one robot. We believe that this type of combined system is necessary for proper laughing behavior, not simply just detecting a laugh and responding to it,” the researchers explain.

While they don’t believe their work is concluded, they believe it’s a promising start on the path towards equipping AI with a better understanding of humor. They next plan to train it on various other laughing functions, such as unshared laughs. They also accept that laughter is one aspect of appearing human, and therefore much more work is required to create truly lifelike machines.

“Robots should actually have a distinct character, and we think that they can show this through their conversational behaviors, such as laughing, eye gaze, gestures and speaking style,” they conclude. “We do not think this is an easy problem at all, and it may well take more than 10 to 20 years before we can finally have a casual chat with a robot like we would with a friend.”

Facebooktwitterredditpinterestlinkedinmail