Artificial Intelligence Is Learning How to Exploit Human Psychology for Profit
Empathy is widely praised as a good thing. But it also has its dark sides: Empathy can be manipulated and it leads people to unthinkingly take sides in conflicts. Add robots to this mix, and the potential for things to go wrong multiplies.
Give robots the capacity to appear empathetic, and the potential for trouble is even greater.
The robot may appeal to you, a supposedly neutral third party, to help it to persuade the frustrated customer to accept the charge. It might say: “Please trust me, sir. I am a robot and programmed not to lie.”
You might be skeptical that humans would empathize with a robot. Social robotics has already begun to explore this question. And experiments suggest that children will side with robots against people when they perceive that the robots are being mistreated.
a study conducted at Harvard demonstrated that students were willing to help a robot enter secured residential areas simply because it asked to be let in, raising questions about the potential dangers posed by the human tendency to respect a request from a machine that needs help.
Robots will provoke empathy in situations of conflict. They will draw humans to their side and will learn to pick up on the signals that work.
Bystander support will then mean that robots can accomplish what they are programmed to accomplish—whether that is calming down customers, or redirecting attention, or marketing products, or isolating competitors. Or selling propaganda and manipulating opinions.
The robots will not shed tears, but may use various strategies to make the other (human) side appear overtly emotional and irrational. This may also include deliberately infuriating the other side.
When people imagine empathy by machines, they often think about selfless robot nurses and robot suicide helplines, or perhaps also robot sex. In all of these, machines seem to be in the service of the human. However, the hidden aspects of robot empathy are the commercial interests that will drive its development. Whose interests will dominate when learning machines can outwit not only their customers but also their owners?