most of us Can distinguish between belly-breaking laughter in response to a cat video and a weak chuckle after a coworker’s cliched joke. But can robots?
at recent days study Published in the magazine prospects in robotics and artificial intelligence, The researchers designed a new AI system to detect people’s laughter, determine whether they should laugh in response, and choose the appropriate type of laughter for context.
This new design may help spark conversations between people and robots in an increasingly digital world.
“I hope we can reinforce the idea that laughter should be an essential part of any conversational bot,” says the study author. Devesh Lala, a researcher studying chatbots at Kyoto University in Japan. “We have proposed the idea of shared laughter as a way to attack this issue.”
Here’s the background – The past decade has brought awful and highly realistic AI-powered robots that can communicate relatively easily with people. And every new tool seems to sink deeper into the uncanny valley.
Take, for example, Sophia: The Human Machine created by Hong Kong-based Hanson Robotics in 2016. Since then she has served as a UN ambassador and spoken at conferences around the world (including an infamous appearance at SXSW where claimed will “destroy humans”).
There is also an upcoming Tesla Optimus that the company will be doing preview On September 30th. Musk believes that robots will eventually mow the lawn, care for the elderly, and act as friends…as well as sex partners.
“Nowadays, conversational agents, interactive avatars, etc. are more than just tools… they have become much more than we used to have when interacting with computers,” Özge Nilay Yalçına cognitive scientist at Simon Fraser University in Canada.
Despite recent discoveries, scientists have struggled to make the robots laugh — a crucial step that some experts see could foster a real, emotional connection between humans and hominins.
Lala says previous work has mostly been aimed at designing robots that can detect people’s laughter. But he and his team wanted to take things a step further. “If you can do that, you can simply create a common laughter system that laughs when a person laughs,” he says.
what’s new – Kyoto University scientists have created what they call a shared laughter system that they hope to eventually program into talking robots.
Here’s how it works: When someone laughs, neural networks pick up the sound. Then, a series of models that categorize the data decide whether to smile in response and, if so, choose the appropriate type of laughter to respond.
More specifically, the system can choose between a “social” or “fun” laugh, categories based on previous studies that rated our chuckles.
Social laughter – which most people unfortunately know – fills silence rather than expressing true delight, while we use cheerful variety in response to something truly funny (such as goodness. DALL-E Mini Meme).
why does it matter – This latest work is just a last-ditch effort to make robots appear more empathetic and help them form meaningful relationships with humans. After all, talking robots could one day look after our elderly relatives and follow us around our homes, along with other particularly intimate applications.
Some experts claim that computers offer only superficial empathy, and that it is a task doomed from the start. “Now machines are not content to show that they are smart; they pretend to care about our love lives and our children,” Wrote Sherry Turklea sociologist and psychologist at the Massachusetts Institute of Technology, in her 2021 diary Sympathy Diary.
But according to Lala, the AI laughter system is not meant to duplicate the real thing. “We don’t claim that our robot can show real empathy, as this requires them to understand human nature as such,” he says. “During the interaction, the robot is just trying to simulate what an empathetic human would do.”
In fact, Lala warns that the AI system does not necessarily detect humor. “She might do it unintentionally but she can’t tell if what you’re saying is funny, especially if you’re not laughing at yourself,” he explains. “While that would be great if it could be achieved, we don’t want to overstate what we are doing.”
He still believes that simulation can help robots better benefit humans, especially those who struggle with social isolation, such as the elderly at home. Even if man does not really understand his human companion, he says, it helps in someone’s presence (or rather, Something) there to listen.
“One of the ethical issues is whether or not we want bots to lie and to openly say things like ‘I understand what you’re going through,’ and that’s something we have to think about,” he says.
What did they do – The researchers trained the AI system using data collected from a speed dating experiment conducted between Kyoto University students and… Erica, a person-like Android system designed by a lab to study human-robot interaction. In this scenario, Erica is voiced by amateur actresses seated in another room.
They examined the audio from these sessions and identified more than 3,000 individual laughs, categorizing them into the social and fun categories. The team also noticed when the actress-managed Erica copied human laughs.
To test the final product, the team hired more than 30 people to listen to an audio recording of the AI system while chatting with people, including study author Koji Inoue.
They performed three different conditions: the main common laughter system, which is a system that does not laugh at all, and a less subtle system that always responds to human laughter with only social laughter. After listening, crowdsourcing participants rated the shared laughter system as the highest ever in terms of “empathy, naturalness, human similarity, and understanding,” according to the study.
What’s Next – Now, Lala and his colleagues are currently working on integrating the AI system into the ERICA android system, along with other chatbots they are working on in their lab.
These future studies will be important to establish whether or not this system actually works Khit b. Trong, an algorithmic linguist at the University of Twente in the Netherlands who studies laughter. After all, she points out, the concept can go south very quickly.
“if [the robot is] Laughing at the wrong moment, it will ruin your relationship with the agent.” “It is very difficult to create a laughing factor because the cost of error is so high.”
Looking into the future, researchers believe that robots could theoretically laugh on their own — not just when people ask for them. But that will require AI to really pick up on the humor. Researchers working in natural language processing, an area of artificial intelligence focused on understanding how people write and speak, are now trying to do so.
Yalcin wonders how we can explain, say, absurd humor like that in Monty Python for PC.
“It’s not very straightforward. It requires you to create a data set that contains all the cultural, social, and personal interactions and all the complexities of the world,” she says. “But I would say this is a good study as a first step.”
Until we get to this point (if we’ve done it before), the new AI system may prove to be very useful as is.
Based on current technological advances, Lala says it could take up to two decades for humans to enjoy a well-intentioned conversation with bots. Besides the element of laughter, machines also need to improve skills such as eye contact, taking turns speaking, and showing interest in the other speaker.
“I think the progress is gradual, but we still have a bit to do before we can all talk to a robot being a problem solved,” he says.
#robots #acquired #frightening #skill