Re: AUI - Artificial Unintelligence

On 21 Mar 2006 11:36:15 -0800, "JGCASEY" <jgkjcasey@xxxxxxxxxxxx> in
comp.ai.philosophy wrote:

Anssi Hyytiainen wrote:

Let's start with a partial definition of intelligence;

Lester Zick wrote:

Let's not. The whole idea here, Anssi, was to develop
a separate theory of unintelligence precisely because
we can't reach consensus on intelligence.


Isn't that like defining a "cat" by what isn't a "cat"?

Sure. The difficulty here is that we don't seem to have made any
progress in defining intelligence. So I thought it might be easier to
approach the subject the other way around so we can state definitely
what each of us considers unintelligent.

The length of the definition would be enormous or it
would be a simple negation of the definition of "cat"
which means you would have to define "cat" first.

The problem is we haven't made any progress defining intelligence. So
we might just as well try to come to some consensus on what it means
to be unintelligent so we can find out just how big a problem we face
in defining intelligence.

I don't think "intelligence" is that hard to recognize.

Well you don't think so, John. None of us thinks so when it comes to
what each of us means by the term. We just don't seem to have any
common ground.

Essentially an intelligent system finds its way to some
goal, some desired state, be that finding the solution
to an equation, finding some literal location, finding
a way to checkmate in a game of chess, finding a way
to get money and so on ...

Sure that's your definition for intelligence and would presumably
represent what you would try to mechanize artificially. Others might
try something else. And in the end all we'd wind up with would be
different definitions which we already knew to beign with.

The problem arises when we confuse being intelligent
with sentience and the problem arises when we don't know
the inner mechanics that allows the system to achieve
those goals.

Well the problem arises when people draft definitions for intelligence
which aren't universally exhaustive and mechanically demonstrable.
Then it's hopeless from the getgo. So I just suggest we concentrate
instead on what each of us considers unintelligent in the hope that a
consensus might emerge that way.

~v~~

.