Role-Based Ethics for Decision-Maker Alignment

Role-based ethics posits that AI decision-makers must perform tasks typically conducted in professional roles with a level of competence at least equivalent to that of their human counterparts. In many professional domains, rules, standards, and guidelines provide an optimal answer to decision problems in a static context. However, in dynamic scenarios, these guidelines may not cover all possible conditions, or the optimal solution may be unattainable due to operational constraints. This may lead human experts to make choices that are contextually valid according to prior cases but deviate from textbook solutions. This paper introduces a framework for validating the outputs of an Algorithmic Decision Maker (ADM) designed to emulate human decision-making in professional roles by evaluating adherence to formal rules while accounting for justified deviations represented in analogous past decisions. We demonstrate its application in military medical triage using the Tactical Combat Casualty Care (TCCC) guidelines and related references as sources of professional standards.