Subject:
|
Re: Design
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Mon, 5 Dec 2005 18:48:55 GMT
|
Original-From:
|
dan miller <danbmil99@yahoo%StopSpam%.com>
|
Viewed:
|
1343 times
|
| |
|
|
--- Mr S <szinn_the1@yahoo.com> wrote:
> Dan,
> I wouldn't have argued it quite that way, but then I
> wouldn't have been so eloquent. On the human mind and
> simulation (mimicry) of it, I have three observations:
>
> Intelligent behavior has a goal (easy to mimic)
> Intelligence has an agenda (not easy to mimic)
> Intelligence has an attention span (I have never even
> heard of anyone trying to mimic this)
>
> In any test of intelligence or test for intelligence
> that I have heard of, none pass the last two points.
> Yet, somehow, even a 4 month old baby is able to have
> both an agenda and an attention span even though it
> has failed to learn anything useful to the parents
> thus far, and in fact, would not be able to pass the
> Turing test...
I don't see any problem with programming an agenda, or an attention span.
Are you saying these are intrinsically hard things to do with a computer?
... Yes, I'm saying that
> intelligent entities are born with their intelligence
> in tact. It is not learned, only better defined with
> learning. This has led me to the conclusion that
> intelligence is not a learning process, but the
> definition of the machine that is doing the
> learning... Learning is a by-product of intelligence,
> not the creator of it.
I'll go with that. A dog in a human family grows up with dog intelligence,
not human. Obviously the brain is set up in just the right way to extract
certain types of information from the environment. I long ago abandoned the
"tabla raza" concept of AI (as have pretty much all serious researchers in
the field).
>
> As for the real world and game worlds differing
> because of the predefined definitions of objects,
> walls etc. I have to say that the real world is
> predefined, just a much larger, more complex set of
> definitions. For instance; A privacy fence is a
> geometric object with associated properties. It is
> also a wooden object. It is an object that has
> relative motion. It is also often an object that has
> the properties of many smaller wooden and/or metal
> geometric objects. A fence, on the other hand, can
> have so many more differing properties, such as a
> chain link fence, a white picket fence, and electric
> fence, a barb-wire fence, a chicken wire fence etc....
> and all of the associated properties of each fence. It
> would overwhelm that game programmer to include so
> many possible properties for something that is
> represented usefully by a single set of geometric
> properties.
I agree with all that. My point was that the breakdown of the world into
objects, modulated through limited sensory data such as vision, is a hard
problem. Typical game programs just bypass that problem, going straight to
the orderly heirarchy of objects the programmer started with. They cheat on
a large scale, from the POV of a realistic simulation. It works because the
actors (the game characters) seem to us to be doing it the way we do. The
end result is the same, from the player's perspective.
>
> The real world can indeed be broken down into basic
> geometric shapes and properties... it just takes a
> great deal of memory and recognition skills to do so
> successfully. Not even all humans are able to do this
> successfully.
Good point. This is, to me, the next step of robotics/AI. Before we can
worry about intention, goals, attention, and the big bugaboo, consciousness,
we have to be able to deal with the world as well as a lizard or frog.
(note however there is quite a bit of evidence that other creatures _do_not_
necessarily break the world down in anything like the way we do. In
general, animals appear to be equipped with the fastest, simplest set of
algorithms necessary to survive.)
>
> It will take a great deal of work, but defining a
> computing system (not device) that is able to function
> as intelligence is STILL a hugely complex goal, one
> that remains out of sight.
?? not clear what your position is on this. It's a complex goal, it will
take a great deal of work. The thread is really about whether the academic
research community is going down the right path, and achieving anything of
significance. I say it is; the problems are very hard, but one by one they
are becoming amenable to analysis and solution, at least in specific
domains. The crux of my argument is that this is real progress, not just
thought experiments in toy worlds.
-dbm
>
>
> --- dan miller <danbmil99@yahoo.com> wrote:
>
> > --- steve <sjbaker1@airmail.net> wrote:
> > ...
> > > Searle's own counter to the argument that the
> > > man+the rulebook is an intelligent system is that
> > > the man could memorize the rulebook and step out
> > > of the room able to speak fluent chinese without
> > > begin able to understand a word of what he was • saying
> > > in response to chinese questioning.
> > >
> > > This would leave use with a strange situation.
> > >
> > > On leaving the room, the man would be able to
> > > behave as a chinese speaker - but he conscious
> > > 'English' self would have no idea what he was
> > > saying.
> > >
> > > It would be like having two completely separate
> > > individuals without one skull. Neither would be
> > > able to access knowledge that the other had.
> > > Neither would be able to speak the other's • language.
> > >
> > > The situation would appear to have complete • symmetry.
> > >
> > Except for the fact that the Chinese guy could only
> > understand one word
> > every million years or so.
> >
> > What Serle (annoyingly, yes!) fails to comprehend is
> > that a human being who
> > miraculously had the ability to emulate a Turing
> > machine, which in turn is
> > running a program able to simulate a human mind, is
> > not the human mind he is
> > simulating.
> >
> > Humans who have had their corpus colossum severed
> > (for severe epilepsy) are
> > effectively two centers of consciousness in one
> > body. Tests have been done
> > where they let the left hand touch a toothbrush, and
> > later identify it from
> > a bunch of objects, but the right hand (which is
> > associated with the left
> > side of the brain, and speech), cannot access the
> > information. This is
> > exactly the state of affairs in Searle's absurd
> > thought experiment. (absurd
> > because of the performance problem of a human
> > memorizing enough rules to run
> > a complex program -- but I'll ignore that for the
> > sake of argument). If I
> > ask something in Chinese of the "wrong" person -- in
> > this case, the
> > english-speaking host -- he will know nothing about
> > it, because he has not
> > learned Chinese. But the Chinese person, who the
> > English person is somehow
> > emulating, can talk Chinese just fine.
> >
> > Here's a concrete example of a similar state of
> > affairs that can exist
> > today: A Macintosh has a PC emulator running in a
> > window. On the (virtual)
> > PC, you open a word processor, edit a document, and
> > save it to disk. Then
> > you open a word processor on the Mac, and attempt to
> > open the file. It
> > won't open. (maybe you can't even find it, or if you
> > can, it's in the wrong
> > format). Hmm, does that say anything about the
> > relative merits of PC's
> > versus Mac's? Does it mean anything at all about
> > their qualitative
> > capabilities? No, it means essentially nothing.
> > The PC could emulate the
> > Mac, and the result would be the same. That's the
> > nature of the beast when
> > it comes to Universal Turing Machines. If Searle
> > (and the legions of
> > philosophy students who actually believe this crap
> > -- I've met a couple)
> > spent some time understanding the Church/Turing
> > thesis, they wouldn't be so
> > confused by a simple sleight-of-hand parlor trick,
> > and the world would be a
> > better place.
> >
> > It's all so silly, because it's just going to
> > happen, and then we can argue
> > about it forever. Or until they get annoyed.
> >
> >
> >
> > __________________________________________
> > Yahoo! DSL – Something to write home about.
> > Just $16.99/mo. or less.
> > dsl.yahoo.com
> >
> >
>
>
>
>
> __________________________________________
> Yahoo! DSL – Something to write home about.
> Just $16.99/mo. or less.
> dsl.yahoo.com
>
>
__________________________________
Start your day with Yahoo! - Make it your home page!
http://www.yahoo.com/r/hs
|
|
Message is in Reply To:
 | | Re: Design
|
| Dan, I wouldn't have argued it quite that way, but then I wouldn't have been so eloquent. On the human mind and simulation (mimicry) of it, I have three observations: Intelligent behavior has a goal (easy to mimic) Intelligence has an agenda (not (...) (19 years ago, 5-Dec-05, to lugnet.robotics)
|
9 Messages in This Thread:
  
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
