Working in science is awesome but sometimes, often at a party, it can be challenging to explain what it is that I do. I studied psychology and now work with, you know… “robots but social ones” as someone put it. Roughly two groups of people ask about my work and job, 1) those who are very interested in novel technology and psychology and 2) those who are scared this new technology is dangerous and “might steal their job”. I love talking about my job so always 1) explain as much as people find interesting and 2) reassure that robotic overlords won’t be happening soon (and that they probably will have an off-switch). In this post you can read what I might say if you ask me at a party “so you work with robots, right…?”
These robots you talk about, I call them Artificial Social Agents (ASA). So it is a social agent, just like a human, but than artificial. An ASA can be a social robot, a virtual agent on a computer screen or some other interface like your phone. An ASA is a computer controlled entity that can autonomously interact with humans following the social rules of human-human interactions. So basically it is a computer that you can talk to, and that talks back, as if it was another human. All ASAs, whether they are social robots or virtual agents, consist of multiple technical components. These components can be roughly divided in three ‘pillars of tasks’: sense, think, and act. Each technical component in an ASA has a distinct task it performs in order for an ASA to function in a social interaction. ‘Sensing’ components are tasked with processing and interpreting the human’s language and social behaviour signals (for example the user’s mouth corners move up meaning a smile). This information is used to ‘think’ about the behaviour of the user (for example the user liked the joke I just told) and based on this it is decided what is an appropriate response (for example laugh with the user to create rapport). The response behaviour (for example smile and laugh) is displayed by the body of the ASA. The human then responds to this behaviour of the ASA and the circle continues. In fact, humans process information in a similar manner!
OK-cool. So what do I do? Over the last years I worked mostly on the technical components that deal with “thinking” about what to do in a social conversation. So what should the ASA do in response to the situation and the behaviour of the human. Currently I am more interested in understanding the effect that ASAs have on the human who is interacting with the ASA. If after this explanation their eyes haven’t glazed over, they usually start asking the real difficult questions… and then we go from there. Or I just show a little clip where a couple of these ASAs explain how they work: