Make Social Bots More Natural

If emotional computing is based on a variety of technologies and systems, social robotics is even more so.Not only do they have to analyze a user's emotions from multiple cues, as affective computing does, but they also have to be able to respond and interact as a separate entity.In the process, researchers have learned a lot not only about how to design these machines, but also about our own psychology and how to succeed.

For robots to truly interact with humans, they should ideally interact with us in the same way we interact with humans.To do that, they need to be able to identify a person and infer what that person is doing and how that person is doing it.This means accurately assessing not only one's external state, but also one's internal state.

This is known in human psychology as "Theory ofMind" (ToM) and mostly begins to develop between the ages of two and five.The development of this ability allows us to identify and attribute mental states of others that are separate from us and independent of us, including knowledge, beliefs, emotions, needs and intentions, and other states.Creating and using theory of mind on robots and other ARTIFICIAL intelligence is seen as a big challenge in parts of the AI community.

There is no guarantee that this step will eventually be achieved, but if we can assume from a reductionist point of view that the mind can be reduced to the physical properties of the brain, then the goal should eventually be achieved.Under this premise, robotic consciousness could one day be possible.Meanwhile, there's another way theory of mind could make robots more like humans, and that's on the human side.From a human perspective, robots don't really need to have an internal state of being aware of each other like humans do.It just needs to imitate well enough to make us believe that it knows.

There is a very important difference between the two, because the jury is still out on whether machines will ever be truly conscious.And until it does, we'll be happy if it mimics well enough.So why do we want to do this?That's because it's the most realistic social interaction we can have through the interface under existing conditions.Given that we still don't know exactly what consciousness is, how it works, or how it comes about, it will be some time before artificial intelligence becomes conscious, if machines ever become conscious.

With a better understanding of how theory of mind works, we may be able to create the illusion that we are interacting with an other with a "perception" similar to our own.Even if it's known that robots aren't really conscious, it's still good for interaction and communication to be willing to believe it.