Explainable, Normative, and Justified Agency

In this paper, we pose a new challenge for AI researchers – to develop intelligent systems that support justified agency. We illustrate this ability with examples and relate it to two more basic topics that are receiving increased attention – agents that explain their decisions and ones that follow societal norms. In each case, we describe the target abilities, consider design alternatives, note some open questions, and review prior research. After this, we return to justified agency, offering a hypothesis about its relation to explanatory and normative behavior. We conclude by proposing testbeds and experiments to evaluate this empirical claim and encouraging other researchers to contribute to this crucial area.