Mike Tintner wrote:
It's driving me nuts because it's basically simple.
Mike, you are getting the wrong end of the stick so badly that you are
actually placing me on the *opposite* end of the argument to where I am,
and then trying to argue against me! ;-)
Here is the thing that is NOT simple: there are some people who claim
that they can give hard, objective definitions for terms like
'intelligence' 'goal' 'solve' etc .... but if you look carefully at
their supposedly "hard, objective definitions" they do not match up with
everyone's commonsense use of those terms. Their definitions do things
like classifying weird optimization procedures as 'intelligent', even
though they don't look intelligent to you or I, and they sometimes
define intelligence in terms of abstract 'algorithms' that require
infinite computing power, so their actual behavior can never be checked
in the real world.
It is those people I am arguing against. The trouble is, disentangling
their definitions in order to SHOW that they are bankrupt is hard work.
It is not simple.
I don't give a hoot about informal, loose definitions, which are a dime
a dozen anyhow (so your offering, below, although interesting, has
nothing to do with the argument), I only care about attempts to produce
a strictly formal, objective definition that can be cashed out without
any reference to things that are subjective ...... and even then, I only
care about those because I am attacking them, not because I am defending
them.
So please, don't put me on a train bound for the nether end of Derrida's
ass: that is what I have been accusing the other side of doing. ;-)
Part of the reason I am doing this, BTW, is that some of those people
are using their 'definitions' to claim superior scientific status for
their work, and then using that supposedly superior scientific status as
a stick to beat down people who are accusing them of wasting vast
amounts of money pursuing an approach to AI that could well be a
complete waste of time.
So this question has real consequences: 90% of the AI research
person-hours might be getting poured down the toilet because some people
are claiming scientific validity for what they do, when in fact they are
just pissing into the wind. This is no joke.
Richard Loosemore.
My definition covers your point:
"
" If I drop this pencil, does it have
the 'goal' of reaching the floor? If 'goal' is the wrong word here
(and clearly it is), then what exactly is a real goal?"
No - only living creatures (and agents with some kind of mind), have
general drives and the power to set specific goals for them, and then
solve the problems of how to reach those goals.
Inanimate matter can only move in basically straight lines and so can
only be said to have "destinations." Put a forked road or obstacle in
its path and it crashes into the obstruction. It has neither the mental
or physical capacity to solve problems.
Living creatures have "goals" - if you put obstacles in the path to
their goals, they have the capacity to solve the problem both
cognitively and physically of how to get around those obstacles and
still reach their goal.
All problem-solving comes down to that.
If you keep redefining the basic terms, like goal, problem etc. , you
achieve 0 but disappearing up your rectum in company with Derrida and a
million other philosophers. Language can never be precise.
The big deal in terms of AGI for me, is polishing up a little the
definition of the second kind of intelligent problem-solving , i.e.
open-ended problem-solving, which is central to AGI . - and then finding
a few choice examples of what is entailed, with the memorability of the
Turing Test example, but without its impossible vagueness..
----- Original Message ----- From: "Richard Loosemore" <[EMAIL PROTECTED]>
To: <[email protected]>
Sent: Thursday, April 26, 2007 9:58 PM
Subject: Re: [agi] Circular definitions of intelligence
Mike Tintner wrote:
You guys are driving me nuts.
Jumping in at the middle, here goes:
"Intelligence is the capacity to solve problems.
(An intelligent agent solves problems in order to reach its goals)
Problems occur when an agent must select between two or more paths to
reach its goals.
Sorry to hear it's driving you nuts, but....
Jumping in at the middle means you missed the original point,
unfortunately: the original point is whether you can do build a
definition without begging any questions, and the terms 'solve',
'problem', 'agent' and 'goal' in the above definition all require
definitions of their own.
When you really push hard on it, it turns out that these terms cannot
be defined without implicitly leaving it up to an 'intelligence' (i.e.
us) to make a judgment call about what constitutes 'solve', 'problem',
'agent' and 'goal'.
Try it: what counts as a 'goal'? If I drop this pencil, does it have
the 'goal' of reaching the floor? If 'goal' is the wrong word here
(and clearly it is), then what exactly is a real goal?
You may have to go back to the beginning of the thread and read
exactly what I was arguing, to know why I said what I did a few hours
ago.
Why is it important? Well, I did say why, too..... :-)
Eric B. Ramsay wrote:
> Several emails ago, both Ben and Richard said they were no longer going
> to continue this argument, yet here they are - still arguing. Will the
> definition of intelligence be able to accomodate this behavior by these
> gentlemen?
Well...... actually I said "Unless you or someone else comes up with
a definition that does not fall into one of these traps, I am not
going to waste any more time arguing the point."
But Ben did try to come up with one, so I continued.
Okay, so at a meta level it was brainless of me to continue ;-)
Richard Loosemore.
-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?&;
--
No virus found in this incoming message.
Checked by AVG Free Edition. Version: 7.5.463 / Virus Database:
269.6.1/776 - Release Date: 25/04/2007 12:19
-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?&;
-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/?member_id=231415&user_secret=fabd7936
