Ethical Healthcare Agents

Summary

We have combined a bottom-up casuistry approach with a top-down implementation of an ethical theory to develop a system that uses machine-learning to abstract relationships between prima facie ethical duties from cases of particular types of ethical dilemmas where ethicists are in agreement as to the correct action. This system has discovered a novel ethical principle that governs decisions in a particular type of dilemma that involves three potentially conflicting prima facie duties. We describe two prototype systems in the domain of healthcare that use this principle, one that advises human beings as to the ethically correct action in specific cases of this type of dilemma and the other that uses this principle to guide its own behavior, making it what we believe may be the first explicit ethical agent.