Humane Anthropomorphic Agents: The Quest for the Outcome Measure ; Position Paper

Artificial intelligence has become an integral part of our daily lives. Today, we engage with intelligent agents at home, on the street, and at work. Rapid advances in technological capabilities make such intelligent agents increasingly human-like. Anthropomorphic agents are characterized by a high degree of socialness, intelligence, and efficiency. They afford many opportunities (e.g., convenience, availability, automation) yet also bring along potential negative impacts on human users, such as uninformed decision making, loss of control, or lack of transparency. Thus, anthropomorphic agents mark a new quality of human-computer interaction that should consider values and ethics in their design process and outcome. However, typical outcomes to measure the quality of an intelligent agent from a user-centric perspective are limited to accessibility, usability, or user experience. In this position paper, we argue that in the design of anthropomorphic agents, we need to go beyond established HCI measures in order to emphasize ethics and values in the digital age. Thus, we propose a new outcome measure called “humaneness” as a foundation for understanding and designing humane anthropomorphic agents.