In recent years, the word “Kansei” has emerged as an important keyword in the field of product development. As high-efficiency and high-performance products become commonplace, sensitivity takes precedence over intelligence in the quest to develop affective products.
Equally, in the realization of natural communication between person and computer, sensitivity-based information processing is a key technique that we cannot do without.
Natural communication is inherently difficult, and we can readily identify many situations in which people fail to achieve mutual understanding.
The following are sentences from a conversation in the movie Back to the Future, in which the main character (Marty McFly) finds himself transported 30 years into the past and is speaking to a younger version of his friend, Dr. Brown (Doc).
In this film, the teenage Marty often says “This is heavy” to express his feelings. His scientist friend, Doc Brown, cannot even understand how he feels something at all.
This is an extreme example, but the use of words to express “ambiguous” human feelings is very complicated. If an intelligent robot is to understand a person with any degree of sensitivity, it must be able to handle the vagueness and individuality of such verbalizations.
“It’s super dangerous isn’t it?”; or “It is super super dangerous!”
These are expressions commonly used by contemporary Japanese schoolgirls. Today, such expressions have become natural in conversations among young people, in ordinary situations that may never become dangerous.
However, an artificially intelligent robot that overheard such a conversation would infer that the speakers were in some kind of danger.
The use of Kansei expression is ever-changing and differs according to age, sex, characteristics or location. If computers or humanoid robots are to communicate naturally (like human beings), they must be able to properly interpret these “ambiguous” expressions and feelings.
To that end, we have devised a subjective Kansei information processing model (SKIP) to advance our understanding of how computers can achieve this kind of human sensitivity and natural communication.