Sensing Emotions, Assembling Humans: Faces, Voices, and Ethics in South Korea’s Digital Companion Research

• In Event: Sensing Robots

Abstract

South Korea’s Ministry of Science, ICT, and Future Planning recently launched a 60 million-dollar research project to develop “autonomous digital companions.” One part of the project is devoted to the research on intelligent robots that will sense emotions of a human being and respond like a sentient entity. By going beyond tossing simple sentences like a ping-pong game, this robot is expected to carry out conversations with proper emotional and ethical reaction, which will make it worthy of the label “digital companion.” In this paper, we will present our effort to follow the engineers who make emotion-sensing robots. Instead of asking, as engineers do, how robots sense emotions, we ask how the engineers define emotion as a “sense-able” quality, how they make sense of a robot as a sentient being, and how they imagine the yet-undefined relationship between sentient robots and emotional humans. In their attempt to render human emotion machine-readable and –distinguishable, the engineers must classify and codify emotions, which requires analyses of facial expressions, voice tones, and speech contents. As these components of emotion are sensed and then assembled to create an emotion profile of a person, ethical concerns arise: How can we ensure ethically sustainable interaction with emotion-sensing robots? Can emotion-sensing robots sense even ethics? How could ethics research on emotion-sensing robots contribute to robotics and sociopolitcal questions? Our research will show that the very idea of sensing emotions with and by machines is entangled with ethical questions about living with and as sentient beings.

Authors

• Heesun Shin, Korea Advanced Institute of Science and Technology

• Hanbyul Jeong, KAIST