Conclusion

This book presents a concise introduction to recent research on human-aware decision-making, particularly ones focused on the generation of behavior that a human would find explainable or deceptive. Human-aware AI or HAAI techniques are characterized by the acknowledgment that for automated agents to successfully interact with humans, they need to explicitly take into account the human’s expectations about the agent. In particular, we look at how for the robot to successfully work with humans, it needs to not only take into account its model M^R, which encodes the robot’s beliefs about the task and their capabilities but also take into account the human’s expectation of the robot model $$M:h^R$$ which captures what the human believes the task to be and what the robot is capable of. It is the model $$M:h^R$$ that determines what the human would expect the robot to do and as such if the robot expects to adhere to or influence the human expectations, it needs to take into account this model. Additionally, this book also introduces three classes of interpretability measures, which capture certain desirable properties of robot behavior. Specifically, we introduce the measures Explicability, Legibility, and Predictability.