Communicating with robots: What we do wrong and what we do right in artificial social intelligence, and what we need to do better

Abstract

Artificial agents, be they virtual agents, or physically embodied devices, such as robots, often require interaction and communication with humans. In addition to the challenge of analyzing human interaction, communication researchers, psychologists, and others are now confronted with new paradigms. Human behavior is not only analyzed, but machines synthesize behavior. Communication is not just observed, but algorithms are employed (in real time) to react to such behaviors. This is a time where social intelligence needs to be put into code and hardware. We discuss the challenges and pitfalls regarding the interaction of humans and machines with a view to (artificial) social intelligence and a time of challenging interdisciplinary research. Concrete examples of such research will be presented and lacunae in empirical data will be pointed out.