Re: [Perl-AI] distributed intelligence
> Hands up who thinks we can build a global web mind with what we've got
> right now.
It's a pretty vague question. Do you mean a mind on a level with human
powers? I say no. Do you mean a mind that is very interesting? We can do
that on our desktops. Can you be more specific about the capabilities of the
mind we're voting on?
I'll say right now, if your proposal includes any kind of significant
handling of memories, such as holding a topic of conversation with something
much more sophisticated than eliza-type keyword mapping, my hand is down,
down, down.
In my (foolish) opinion, the issue of memory strikes at the heart of
intelligence, and is far, far behind the current state of the art of nn, ga,
and fuzzy logic. We don't even have any meaningful models to experiment
with.
Please correct me. I would love to be wrong.
Zack
p.s.: I think a global neural network comprised of participating client
machines, and requiring no central server (!), is currently possible and
would be quite fascinating.
I'm working on a system that implements this. I've been calling it aidist.
It creates a global distributed network in which there is no need for a
central server, and each participating node acts as an input node, a hidden
node, and an output node. The brain organizes itself into indistinct
clusters of compatible and mutually enhancing activities, based on the
desires of the participants.
I think of aidist more as a distributed computer than a distributed mind,
although ai, or even linguistic communication between the global structure
and individual participants, are conceivable.
28
> Dec.
> --
> +------+
> ==>| GIGO ==>
> +------+
>
Reply to: