Moral Decision Making Frameworks for Artificial Intelligence

• Vincent Conitzer, Walter Sinnott- Armstrong, Jana Schaich Borg, Yuan Deng, and Max Kramer.
• developing a general ethical decision making framework for AI based on game theory and machine learning
• For the game theory based framework, the authors suggest the extensive form (a generalization of game Trees) as a foundation scheme to represent dilemmas
• current extensive form does not account for protected values in which an action can be treated as unethical regardless of its consequence
• extend the extensive form representation with passive actions for agents to select in order to be ethical
• machine learning based ethical decision-making
• classify whether a given action under a given scenario is morally right or wrong
• The main challenge in machine learning based moral decision-making is to design a generalizable representation of ethical dilemmas
• Game theory and machine learning can be combined into one framework in which game theoretic analysis of ethics is used as a feature to train machine learning approaches