Speaking as a software developer, I find it difficult to imagine that artificial intelligence could NOT be developed. Furthermore, I find it impossible to believe that if we DO develop AI, that it CANNOT be 'superior' to it's creators. I've encountered a number of difficult situations in my times, and I have never encountered one that hasn't fallen to the twin tools of divison and abstraction, either in my mind, or the mind of a person or persons brighter than me... More difficult problems simply require a more ingenious choice of the locations to cut with division, and the concepts to wrap in abstraction.
Let us apply the concept of problem-domain division to the problem of artifical intelligence:
Theoretically, you need only a completly functional electronic model of a human neuron, and a complete map of a basic human brain. Let us suppose we have the chemical and physical description of a neuron, and a willing volunteer (Doug) that submits to a invasive, perhaps destructive, technique, wherein a set of nano-machines populates the subjects neurons and synapses. The machines reports back on the fundamental structure of the underlying synapses...
Feeding this neural structure into the brain, using our billions upon billions of simulated neurons, we produce a copy of Doug... Now, this may or may not be a self-aware entity, that is a philosophical debate... But I would argue that we have, by any real literal breakdown of the term, created an ARTIFICAL intelligence. It would speak, act and respond as Doug would.
Yes, at this point, it is no better or worse than Doug. (This, of course is assuming that we cannot simulate the neurons physical interactions at twice the speed of reality, or faster...)
However, let us add a unit to this brain, assuming we have some fundamental understand this neural map... This unit will detect a particular neural pattern, which we might train the 'new' Doug to project, and allow Doug to then "think" two X digit numbers, which would then be multiplied and returned to Doug... Doug has now gained savant-like multiplicitive abilities.
This, I think, being the old Doug, with a single newfound ability, is superior to the old, if even in a small way. However, my intuition says that if a small improvment can be made, there would be little reason that larger, more drastic impovements could not be made...
Slashdot Mirror
Speaking as a software developer, I find it difficult to imagine that artificial intelligence could NOT be developed. Furthermore, I find it impossible to believe that if we DO develop AI, that it CANNOT be 'superior' to it's creators. I've encountered a number of difficult situations in my times, and I have never encountered one that hasn't fallen to the twin tools of divison and abstraction, either in my mind, or the mind of a person or persons brighter than me... More difficult problems simply require a more ingenious choice of the locations to cut with division, and the concepts to wrap in abstraction.
Let us apply the concept of problem-domain division to the problem of artifical intelligence:
Theoretically, you need only a completly functional electronic model of a human neuron, and a complete map of a basic human brain. Let us suppose we have the chemical and physical description of a neuron, and a willing volunteer (Doug) that submits to a invasive, perhaps destructive, technique, wherein a set of nano-machines populates the subjects neurons and synapses. The machines reports back on the fundamental structure of the underlying synapses...
Feeding this neural structure into the brain, using our billions upon billions of simulated neurons, we produce a copy of Doug... Now, this may or may not be a self-aware entity, that is a philosophical debate... But I would argue that we have, by any real literal breakdown of the term, created an ARTIFICAL intelligence. It would speak, act and respond as Doug would.
Yes, at this point, it is no better or worse than Doug. (This, of course is assuming that we cannot simulate the neurons physical interactions at twice the speed of reality, or faster...)
However, let us add a unit to this brain, assuming we have some fundamental understand this neural map... This unit will detect a particular neural pattern, which we might train the 'new' Doug to project, and allow Doug to then "think" two X digit numbers, which would then be multiplied and returned to Doug... Doug has now gained savant-like multiplicitive abilities.
This, I think, being the old Doug, with a single newfound ability, is superior to the old, if even in a small way. However, my intuition says that if a small improvment can be made, there would be little reason that larger, more drastic impovements could not be made...
Anyhow, Rambling, Goodnight!