MoralDM

  • enables an agent to resolve ethical dilemmas by leveraging on two mechanisms
    1. first-principles reasoning, which makes decisions based on well-established ethical rules (e.g., protected values); and 2) analogical reasoning, which compares a given scenario to past resolved similar cases to aid decision-making.
  • the exhaustive comparison approach by MoralDM is expected to become computationally intractable
  • [Blass and Forbus, 2015], MoralDM is extended with structure mapping which trims the search space by computing the correspondences, candidate inferences and similarity scores between cases to improve the efficiency of analogical generalization