Polite laughter, bursts of laughter, sudden bursts of laughter… We could present the 50 tones of laughter that human beings have been capable of developing. Now it’s up to artificial intelligence to acquire them. This is one of the works of Kyoto University, which hopes to make humanoid robots more empathetic just by aiming for laughter.
In a study published on the Frontiers site, researcher Koji Inoue explains that he is committed to teaching the art of conversational laughter, which is the practice of having the right laugh, at the right time. The stakes are high, as humanoid robots will only be tolerated and truly accepted if they leave the famous “valley of the strange.”
This concept, theorized by another Japanese roboticist, Masahiro Mori, holds that the more the robot resembles a human being, the more disturbing or even monstrous its imperfections seem to us. Therefore, a giggle at the wrong time would not help the robot to be accepted.
For this reason, the Kyoto University team has launched a series of speed dating between students and a robot, called Erica, so that artificial intelligence stores as much data about laughter, whether natural or a little forced.
After this work, the team put Erica’s new sense of humor to the test through four dialogues, which were then broadcast to 130 volunteers. Compared to the same scenes but without laughs, the scenes were considered to be more emphatic, more natural.
Beyond laughter, robotics researchers are also working on generating emotion. And that goes through artificial intelligence but also through the subtlety of human facial expressions. In this regard, the ultra-realistic Ameca robot, again built by a Japanese team, reaches amazing degrees of finesse.
Than to reduce, always a little more, the depth of this dying valley of the strange…
Source: BFM TV
