Understanding individual and group ethical problem solving: A computational ethics approach.

摘要

This dissertation begins by integrating the results of theorizing in several disparate disciplines: moral philosophy, moral psychology, business ethics, group dynamics, personality traits and communication tendencies, operations research, cognitive science and artificial intelligence. We begin by reviewing ten normative and twenty-three descriptive ethics theories. We review the Theory of Planned Behavior from psychology and show how it can inform an ethical-problem-solving-focused implementation of the Belief-Desire-Intention model of artificial intelligence. We summarize work-to-date on computational models of groups as well as those of ethics. We report research that shows that there is promise for using information technology to support ethical problem solving. We report research that found sixty-eight relationships among value types, ethical ideology dimensions, reasoning heuristics, and decision components chosen when an ethical problem was solved. We share a design for a software agent that can mimic the ethical problem solving of an individual. We describe the design, development, verification and validation of a computational model of a group of individuals solving an ethical problem. We use this model to run virtual experiments based upon varying group size, group composition, and rules regarding when a group member can decide whether a proposed solution component should be included within the solution. We report several findings from these virtual experiments, at least two of which indicate that in some group ethical problem solving situations nonintuitive results may occur. We share four prospective roles of an ethical problem solving decision aid that could use computational models similar to those described. Finally, we indicate plans for further research.; Contributions of this research include (1) showing that in certain circumstances, information technology can be used to aid in ethical problem solving, (2) showing that ethical problem solving can be mimicked via the symbolic paradigm of artificial intelligence/cognitive science, (3) providing a template which can be used to create computational models of ethics, and (4) identifying specific relationships between personal value types or ethical ideology dimensions and reasoning heuristics used during ethical problem solving.