Some scientists are working on a much more serious project than it seems: robots that laugh

Do they tell you a joke at the bar counter? You laugh. Are you on the bus and remember a funny anecdote? You laugh. Do you see someone slip on the street like Bambi on ice? Are you new to the office and your nerves eat you up? Of course, you laugh. You laugh if the boss shares a more or less funny anecdote, even if it’s less than more. And it is December 22th and the children of San Idelfonso sing the number of your tenth, what are you doing? The same: you laugh. Wow if you laugh.

You laugh in each and every one of those cases, although, of course, it has little to do with the lazy giggle with which you hide your first days in the office with the laughter of euphoria at knowing you, suddenly, the richest person in your city. The same if the girl you like is looking at you.

You know it and you understand it. A machine, not so much. If not, let them tell C-3PO.

Knowing when to laugh and which laugh is the most appropriate in each circumstance is not an easy task. Especially if we are surrounded by people. It is necessary to properly interpret the context and, in case we are talking to one or more people, the conversation itself. We have it so internalized that we forget the number of nuances that influence it, but no, it is not simple.

A team from the Kyoto University has been set the ambitious goal that, in their own way, robots can laugh when and how. To achieve this, they have taught an artificial intelligence (AI) system “the art of laughter” during conversations with humans, a far from easy task for which they collected data from more than 80 talks in which students and the AI ​​itself participated. a system that Japanese scientists have baptized “Erica”.

Don’t laugh, which is worse

The experience served to accumulate information about solitary laughs – that funny joke that you remember suddenly, silently – and those that arise when interacting with other people. Among the latter, they differentiated between laughter of joy and others caused by less obvious reasons, such as embarrassment, nerves or a simple matter of courtesy. With such background, the Japanese team dedicated themselves to training Erica, basically teaching her two keys: when to laugh and how to do it.

“Our biggest challenge was identifying the real cases of shared laughter. It’s not easy because, as you know, most laughs aren’t really shared at all.” explains Dr. Koji Inouefrom Kyoto University to Guardian. The task seems anything but fun: Inoue and her colleagues they had to categorize the laughschoosing which ones were worth for their study.

After all, one cannot always respond to a smile with another smile. Imagine that the person in front of you laughs out of embarrassment. If you imitate him, he will think that you are making fun of him.

With their “recipe” for Erica’s sense of humor ready, the scientists decided to put it to the test. They designed four short dialogues, ran the new laugh-share algorithm, and then showed the results to about 130 volunteers who evaluated them according to questions such as their naturalness, if they reflected understanding or resembled a human reaction.

The key, detailed the researchers, who have collected the entire process and their conclusions in Frontiers in Robotics and AI, is always the same: identify well the interlocutor’s laughter to which we can respond and select the best way to do it. “The prediction and selection showed higher scores than a random model”, explain its authors, who also detail that the level of detection was also high. His model, in short, conveyed a greater sense of empathy than others in which, quite simply, you always respond with social laughter.

The million dollar question at this point is: So that? What exactly are scientists looking for? Why are we interested in teaching robots the art of laughing when they are chatting with people?

The answer is simple: design more natural AI, with a greater capacity for interaction. “We believe that one of the important functions of conversational AI is empathy —points out Dr. Koji Inoueone of those responsible for the study, to the British newspaper—, so we decided that one way in which a robot can empathize with users is by sharing their laughter”.

dialogue systems. must be able to express empathy to achieve a natural interaction with humans. However, generating laughs requires a high level of dialogue understanding. Implement laughter in existing systems, such as conversational robots, it has been a challenge”.

Does that mean we’re a little closer to sharing jokes with a robot? The Japanese team believes that laughter could be used to design robots with a peculiar character. Of course, Inoue assumes that it will be decades before we can chat with a robot as we would with a friend. Others simply emphasize that an AI will never understand us. Not us, not the laughter itself.

Cover image | Maximalfocus (Unsplash)