Main menu


It's humorously difficult to train a robot to laugh

featured image

Laughing robots may look ominous, but if you laugh at the right moment during a conversation, does it make you less spooky? That’s a theory scientists are testing these days. A group of researchers from Kyoto University in Japan mocked up a laughing robot called Erika powered by his speech-focused AI system.

They reasoned that since laughter is a normal part of human conversation, it might be helpful to see how people respond to chatty robots that can share laughter.Their findings were published last week in the journal The Forefront of Robotics and AI.

Artificial intelligence is good at logic, but laugh? Not really. First, researchers recognize that there are many different reasons why people laugh, which makes matters more complicated. To make the AI ​​system easier, they generalized laughter into his two categories. Shared social laughter when the AI ​​laughs in response to human laughter, and solo hilarious laughter when the robot laughs in response to a subject or while speaking.

The research team trained an AI model on when and how to laugh by allowing them to participate in speed dating with male college students. Erica was remotely controlled by a female actress who spoke into a microphone and controlled body movements such as nodding her head and other gestures.

Chats lasted 10-15 minutes and data was obtained from 82 conversations. The researchers used microphones and cameras to record conversations and annotate them based on when social and solitary laughter in humans occurred and how these laughters differed. He then used this data to train an AI system to learn when to laugh and what type of laughter to use. We then applied the shared laughter algorithm to existing conversational software and listened to 130 volunteers to assess how well the robot simulated empathy, understanding, and humanness.

Overall, in situations where sharing a laugh was appropriate, Erica and her algorithms did a good job of convincing people that they were paying attention to what was being said. It had its drawbacks and limitations. Erica was good at responding to laughter, but she didn’t know when to laugh. In discussion, the researchers wrote that it may be because it’s easier to learn to respond to prompts than to actually understand what makes a conversation interesting.

Whether Erica can truly understand human humor is just one part of a larger project roboticists and engineers are working on. It’s about giving robots social skills. Since 2017, scientists have been working on how to make robots laugh convincingly (with tech giants such as Microsoft, IBM, and Meta also interested). A month ago, an Italian engineer unveiled a bartending girlfriend robot that can make small talk (unfortunately, it’s been shelved for the time being due to privacy concerns). The idea is that by giving robots facial expressions, body language, speech, and the ability to understand and respond to people’s demands, they will become more engaging and improve their daily interactions.

Ultimately, though, there can be a slippery slope from social interactions that feel natural to uncanny valley scenarios. There are also ethical concerns with robots so implausible. Nevertheless, there are practical reasons for continuing to work in this field. Giving talking robots appropriate human-like features to make them less spooky and more friendly, experts say, is especially useful when integrating robots into healthcare, hospitality, or other services. I think. service-oriented industry.