Hiya! It is incredibly interesting to me! Before I answer your questions, let me just say that I see us as highly complex chemical machines. Our experience of free will is an illusion produced by a highly complicated neural network. Given this, emotion is part of the neural network, and therefore can be replicated into any media capable of handling a neural network of the complexity of the human brain, including future computers.
Sorry, I was going to try to keep God out of this thread, but I find with my viewpoint |I can't, so I hope you will bear with me.
It's interesting what you say about free will, as I think free will is commonly misunderstood...I've had to think a lot about that since becoming a Christian. It sees to me that people haven't really got free will, at least in some senses...trouble is I'd have to go into an awful lot of explanation to put over what I mean, and of course, it would be from my perspective, as opposed to, or even maybe as well as our being complex chemical machines.
epTo emulate human intelligence in a computer would require a greater understanding of what makes us tick than what we have now. AI research and neurobiology evolve hand in hand, with computer modeling of neural networks providing valuable data to biologists while experiments on organic neural networks (brains) provides data that makes our computer simulations more accurate.
In terms of hard system requirements, the blue brain project is capable of emulating between 10,000 to 100,000 neurons and their associated synaptic connections within the neocortical column of a rat brain using a single bluegene/L super computer. With around 100 billion neurons in the human brain, this is a lot of really expensive super computers. The number of super computers goes up further, however, when you realize that the connection density has a direct corelation on runtime, meaning that to ke your virtual brain 'thinking' at the same speed we think would require an exponential increase in processing power at a rate dependent upon the average number of synaptic connections per neuron, so we're talking a *LOT* of computational power.
To detail the origin of this problem, you need to understand that a computer is a serial processing device (more or less) whereas the brain is not. A processor does one thing at a time, really fast. RISC processors are the epitome of this, and CISC processors sort of alleviate this by allowing several functions to be done by a single operand (modern computers fall in between these two extremes). Streamlining the pipe allows a more efficient use of processor resources per job, hyperthreading lets a processor use disparate resources within it for different jobs, but any single resource is dedicated to a job for the time that a given instruction is taking place. Multicore processors are valuable to neural net research because it breaks the serial nature of computer processing, but not by enough to manage very large scale neural networks like a cat's brain.
The development of a completely new processor architecture is needed to really let neural network technology take off, and with the creation of memristors, this is now a real possibility. This new processor will have to be asynchronous between it's constituent components, and have a unique bus architecture that can handle the exponential scaling inherent in neural net communications, a serial bus would turn it into a normal processor, while a true neural bus (where every node connects to every other) is impractical. At this point, it's my opinion that bus architecture is the biggest hindrence to the development of industrial capable solid state neural networks.
...Sorry for the rant *blush*
That's quite all right! I myself can rant on for ages sometimes, if I don't restrain myself
Thanks for the comprehensive explanation...I thought it would be tricky to copy everything our brain does! Our brains are incredible aren't they.
nt
entitBack on topic, I don't believe in a soul, so in my opinion an AI android would be equal to me as a thinking, intelligey. If the AI were created with emotions, would you feel it's okay to treat it as below you?
I think religion has served a vital purpose in human society, and this is why it's so prevalent now. It is not, however, truth. External stimulus has a direct impact upon our brain structures, which is why gay men having female-like brains is not indicative of a gay gene, and why religious experience having a direct impact on the brain is not proof of religion's validity. Both these are causation/corelation errors.
(To note, I do believe that sexual preference has a certain amount of both nature and nurture to it, but touting brain structure as proving nature is a logical fallacy.
)
Yes, that is sort of what I was thinking about, if AI android was or seemed human, how would we relate to it? As you say, physically, we are a bunch of chemicals. Do you reckon an android (am using that term cos of sci fi, don't know if it's what scientists would use) could have personalities?
So do you mean it's our individual brain structure, plus external stimulus, that makes us a certain way, for example religious or whatever, or have I misunderstood you?
[/quote.
Sorry this post is all spaced out, but for some reason, I was having trouble with the quote thingies