While artificial intelligence continues to evolve day by day, the most important factor that separates human and artificial intelligence from each other in its current situation has still not been overcome: Emotions. This factor, which binds humanity to each other, also creates the necessity of having a human translator, especially in literary translations, despite the very advanced translation technologies today.
So, what are the emotional states of robots these days when we talk about humanoid robots? In fact, very little to say no. Because robots only act according to the inputs provided by humans and the ready-made commands we design as a result of these inputs. In Japan, which stands out in robot technology, scientists have started a new study that will take it one step further.
Can robots laugh? On a basic level yes, but as a human no:
Scientists from Kyoto University have created an Android robot called ‘Erica’. The main reason for creating this robot was to ‘teach a robot, artificial intelligence’ to laugh. The robot began to be trained to understand the human perception of ‘humor’ and to laugh in different ways depending on the situation and place.
To teach Erica to laugh, the scientists first developed the ‘shared laughter’ model, which would allow the robot to respond emphatically to human laughter. In this model, three questions were set for the robot to respond appropriately: Did the user laugh? Will Erica laugh? If he is to laugh, with what kind of laughter will he laugh?
Laughter left to Erica’s choice was designated as a polite chuckle to fill the conversation in the absence of humor, and a cheerful laugh to be used in funny situations. In order for the model to be used effectively, the robot established a dialogue with 80 university students. The laughter produced by the participants in these dialogues was categorized as ‘single, social and cheerful’.
After that, Erica’s first-of-its-kind sense of humor was put to the test by sitting down for some conversations with a human. Four different short dialogues were prepared, each of which caused a different laugh. However, the team created two more models for comparison. One of these models was one in which he never laughed while speaking, and the second was that he laughed socially when the person laughed, regardless of the context.
The dialogue established by the robot and its responses were evaluated by more than 130 people under the headings of empathy, spontaneity, humanity and understanding. Erica’s first developed model of laughter achieved much better results than the other two models. But according to scientists, Erica is still limited to social and cheerful laughter and needs many types of training to have natural, human-like conversations.
So much so that the head of the study, Dr. “We don’t think this is an easy problem at all, and it may take more than 10 to 20 years before we can finally have a casual conversation with a robot the way we do with a friend,” says Koji Inoue of robots’ inability to display human emotions. He used his expressions.
The research was published in the journal Frontiers in Robotics and AI.