creating AI by mimicking the brain

Simon Laub silanian at mail.tele.dk
Sat Feb 16 18:20:50 EST 2002


"codeZ" <deevv at hotmail.com> wrote news:a3ks0s$8pv$1 at lust.ihug.co.nz...
> I think AI can be created, based on the internet's infrastructure.
> All you need is 30 billion computers connected s.t. each computer connects
> to 10 000 others, a total of 1,000,000 billion connections. (replace a
> computer by a neuron and you have a human brain). Add to this some simple
> software program to manage the input/output and you have a recipe for AI.
> Scan the brain, note all the interneuron connections, then connect the
> computers over the net in the same way. Program each computer to operate
> just like a neuron. Voila. Make sure you have the kill switches if need
be.

Very well, but the brain is situated in a body. What makes
your "neuron computer" add up to centers that process sight, hearing,
feelings etc. (I.e. without an understanding on how these centers
are created and maintained, how can you be sure that you have
not overlooked some little chemical that does some particular trick
that helps keeping this center ticking).

And how do you arrive at the all important sense of "self", when
you have no reference body. I.e. what is the body of the thing
you have created?
Is it some person/robot walking around somewhere
(without a brain) that you are taking input from and creating a
sense of "self" from ?
Surely this is more complex than just connecting a lot of computers.

>How much modelling was required to sequence DNA? It was done in 2 years.
>Then brain is not mush more complex. The connections are larger than DNA
>nodes, hence easier to map. In practice all you hace to do is X-ray it
slice
>by slice, then construct its computer model and record all connections.
Easy
>as 1 2 3. then run a simple program on all 30 billion computers and have
>them self connect in the same way.

Now, you have the representation of one human mind. But without
the body that goes along with it. What does this mean? What if the person
you scanned is blind - then you won't have any idea on how to get your
visuals
going on this system.
How does your system learn?
How does it react to the chemicals when there are no chemicals in your
computer network brain? How do you feed it input form bodily
sensors?

Again - it is just way more complex than just wiring up
a lot of computers.

-Simon





More information about the Neur-sci mailing list