Embedding Ethical Principles in Collective Decision Support Systems

• Joshua Greene, Francesca Rossi, John Tasioulas, Kristen Brent Venable, and Brian Williams.
• authors envisioned a possible way forward to enable human-agent collectives [Jennings et al., 2014] to make ethical collective decisions
• By imbuing individual agents with ethical decision-making mechanisms (such as those mentioned in the previous section), a population of agents can take on different roles when evaluating choices of action with moral considerations in a given scenario
• Based on a set of initial ethics rules, more complex rules can be acquired gradually through learning.
• Their evaluations, manifested in the form of preferences and limited by feasibility constraints, can be aggregated to reach a collective decision
• need for new forms of preference representation in collective ethical decisionmaking
• potential candidate actions to choose from can vastly outnumber the number of agents involved which is very different from multi-agent voting scenarios.
• candidate actions may not be independent from each other, some of them may share certain features which describe their ethical dilemma situations