Should we program our robots and AI to have empathy, and if so how?

As you’ve seen in the “Open up your mind” section of this online learning experience, there is quite a bit of debate about whether AI and robots should be programmed to have empathy. According to Pascale Fung empathy will make interaction with robots more pleasant for humans. Others feel empathy is not needed in AI, or even that the lack of emotional behavior is a core strength of robots. Maybe robots should just do what they’re told, instead of trying to “helpfully” infer our mental state and change their behavior. There are also concerns about the privacy implications of constantly having your mental state assessed and AI systems misleading people into thinking they are conscious, emotional or generally intelligent. Are these concerns warranted, or are there ways in which empathy can be implemented that avoid these pitfalls? What other problems and solutions are there?