Kevin Prescott of ICT results writes, "I thought you might be
interested in a recent EU-funded robotic research
Growing (FEEL, Interact, eXpress: a Global appRoach to develOpment
With INterdisciplinary Grounding). This oddly named project is an
attempt by European researchers to develop a robot with empathy.
Results story on the project says, "The technology pulls together
research in robotics, adaptive systems, developmental and comparative
psychology, neuroscience and ethology, which is all about human
behaviour." The robot will react to facial expression and emotional
states of nearby humans, and reveals its own state via robotic facial
expressions. It will rely on neural networks for learning capabilities.
The researchers believe making robots more empathetic will lead to greater
acceptance by humans.
I would think that there are various stages of artifical intelligence. (1) Ameoba stage where robots have a couple of sensors to grope around stupidly or home in on light or food (wow worms do calculus - not). (2) Simulated intelligence where responses are given based on preprogrammed criteria. This would be the sort of 16 answers to the 16 most asked questions scenario or a program that doctors may use to diagnose an illness. This seems to be what they are shooting for where it acts like it's empathetic from what it observes and may even be quite convincing. Liza type programs or turing tests often stop at this level and don't get much further. (3) The third level would be where the intelligence I would call no longer artificial or simulated. This is where it could actually decide for itself to be empathetic or not. Actual cognitive reasoning and self awareness or sentience. This level of awareness seems to be most illusive. It would be interesting to see them get to this stage. What is the 4th stage? Transcendence? ;-)
The other thing that's interesting besides intelligence is emotion. Where does intelligence end and motion begin? Perhaps that's one of those gray areas that's not tangible. Is it bio, is it intelligence, is it just behavior mimicry. It seems that they are shooting for the monkey see monkey do approach or the if you see that do this approach. Again, it's not the real true empathy, just the simulated empathy. Humans are usually pretty good at telling the difference between simulated intelligence and real intelligence. They're probably pretty good and telling the difference between real emotion and simulated emotion. I'm afraid that the results of this won't be acceptable unless they strive for the second type, the no-simulated type of empathy or "true" empathy. Again this would seem to be most illusive.