Social Causality and Responsibility: Modeling and Evaluation

摘要: Intelligent virtual agents are typically embedded in a social environment and must reason about cause effect. Social causal reasoning is qualitatively different from physical that underlies most current intelligent systems. Besides causality, the assessments of emphasize epistemic variables including intentions, foreknowledge perceived coercion. Modeling process inferences causality can enrich believability cognitive capabilities agents. In this paper, we present general computational model responsibility, empirically evaluate compare with several other approaches.