The Ethics of Robotic Deception

摘要

The time of robotic deception is rapidly approaching. While there are some individuals trumpeting about the inherent ethical dangers of the approaching robotics revolution (e.g., Joy, 2000; Sharkey, 2008), little concern, until very recently, has been expressed about the potential for robots to deceive human beings. Our working definition of deception (for which there are many) that frames the rest of this discussion is “deception simply is a false communication that tends to benefit the communicator ” (Bond and Robinson, 1988). Research is slowly progressing in this space, with some of the first work developed by Floreano et al (2007) focusing on the evolutionary edge that deceit can provide among an otherwise homogeneous group of robotic agents. This work did not focus on human-robot deceit, however. As an outgrowth of our research in robot-human trust (Wagner and Arkin, 2008), where robots were concerned as to whether or not to trust a human partner rather than the other way around, we considered the dual of trust: deception. As any good conman knows, trust is a precursor for deception, so the transition to this domain seemed natural. We were able to apply the same models of interdependence theory (Kelley and Thibaut,