Lyle Bateman wrote:
>
>> Sorry Bill, I gotta disagree here. Beyond the issue of data size and cpapcity of a
> particular arch, the computer that is my brain operates in fundementally different ways than
> the computer that I'm typing this reply on. My brain is *not* a number cruncher, and does
> such tasks extremely poorly and slowly, if at all.
> There are some calculations which, not
> matter how much time was involved, a human computer could not do - take a look at some of
> the weather data processing going on now, or calculations involving various quantum effects
> over a large system.
True, so without tools to extend our ability there seems to be some
limits in our architecture. What are they? We have limited mathematical
ability, our memory isn't perfect, we aren't totally logical, we rarely
throughly think things through, etc.
You could have proved your point more easily though by pointing out that
a simple calculator can't understand english even if a mike was
attached.
However that's not what we are talking about. We are talking about
binary computers (not calculator or human brains). Binary computers can
emulated interconnecting nodes, are extremely flexible, have a perfect
memory etc. They aren't powerful enough but that isn't necessarily a
long term barrier, esp. once if quantum computing comes online (which
will still be digital computing).
So from the above I see no reason a binary computer can't become
intelligent because of its architecture.
>Conversely, there are things I can do that you cannot teach an
> existing electronic computer to do (such as understand plain English and respond in an
> intelligent (OK, a topical) fashion).
Not yet, that is what AI is all about.
>> The difference, I believe, is architecture. My brain is organized in a fundementally
> different way from any current computer system. It is not a single processor, or even a
> multi-processor machine such as a Cray or and Origin. It is a system of billions
> (trillions?) of nodes which interconnect in many different ways, and it is through the
> interaction of nodes, not the processing of the nodes themselves, that brain function
> occurs, IMO.
Nodes can be simulated on a computer eg. neural networks. I do agree
that the architecture has to be flexible enough and powerful enough, but
it seems to me that while maybe not optimum binary computers are
certianly flexible enough (though no where near powerful enough yet).
Cheers,
Keeva.