RE: Empathy and intelligence- It is possible to be brilliant in some respects without empathy, but it is definitely a handicap not to have it. There are many aspects of intelligence, only some of which are measured by IQ tests. Empathy is one, others are musical and artistic talents and social skills. I question whether it is possible to teach an AI any of these, especially empathy. The latest research I’ve read indicates that the ability to develop empathy is tied to what have been labeled “mirror neurons”. They are missing in people with autism- and, quite possibly, in psychopaths too.
Agreed, and I’d put it at at least 50-50 that it’s outright wrong.
(My understanding is that they’ve had lots of trouble doing studies on this, partly because it’s hard to get lower-functioning autistics to focus on the things that the testers want them to pay attention to, so many of the older tests gave results that were, on further examination, incorrect. The most recent results I’ve heard about say that autistic people are at least as empathetic as neurotypicals, on average; the social problems have more to do with difficulty using information gained via empathy than with the empathy not being there at all. I actually wouldn’t be at all surprised if it was an integration issue similar to the sensory integration issues that are so common.)
Hmmm. I am able to achieve a modicum of empathy with my pet dog. I doubt that this is due to my possession of a neuron that fires when I wag my own tail.
RE: Empathy and intelligence- It is possible to be brilliant in some respects without empathy, but it is definitely a handicap not to have it. There are many aspects of intelligence, only some of which are measured by IQ tests. Empathy is one, others are musical and artistic talents and social skills. I question whether it is possible to teach an AI any of these, especially empathy. The latest research I’ve read indicates that the ability to develop empathy is tied to what have been labeled “mirror neurons”. They are missing in people with autism- and, quite possibly, in psychopaths too.
98% confidence that this is at least a massive oversimplification.
Yep: Mirror neurons seen behaving normally in autism
Agreed, and I’d put it at at least 50-50 that it’s outright wrong.
(My understanding is that they’ve had lots of trouble doing studies on this, partly because it’s hard to get lower-functioning autistics to focus on the things that the testers want them to pay attention to, so many of the older tests gave results that were, on further examination, incorrect. The most recent results I’ve heard about say that autistic people are at least as empathetic as neurotypicals, on average; the social problems have more to do with difficulty using information gained via empathy than with the empathy not being there at all. I actually wouldn’t be at all surprised if it was an integration issue similar to the sensory integration issues that are so common.)
98% seems conservative to me. ;)
When I’ve done calibration checks I’m far more prone to overconfidence than underconfidence, so I nudged it down.
And most of which are not even possessed by humans.
Hmmm. I am able to achieve a modicum of empathy with my pet dog. I doubt that this is due to my possession of a neuron that fires when I wag my own tail.
Supposing that you wanted to go the mirror neuron route for developing empathy in an AI, it would need a virtual body linked to its utility function.
Damned if I know whether this is a reasonable path for AI development, but it would be a very handy premise for science fiction.