From: Jochen.Fromm@t-online.de
>
> You can use large clusters for many purposes, one of which is parallel
> IR that Google company is most interested in. Evolutionary programming
> can be considered a grand challenge; however it must serve some end.
> Building a "Matrix" is probably not a sufficient end in itself.
>
I think building a virtual "Matrix" world is a sufficient end in itself.
This
includes of course the construction of intelligent agents, which are able to
understand and live in this "Matrix". You need several clusters, one for the
simulation of the world, and one for every intelligent agent. The two tasks
of
constructing an external and an internal world are closely related, if you
consider that an intelligent agent should have a world knowledge which is
large enough to build up a representation of the world.
Randall Davis says in an AI Magazine article about thinking
and representation (Volume 19 Number 1 (1998) 91-110) that
"thinking is not simply the decontextualized manipulation
of abstract symbols, powerful though that may be. Instead,
some significant part of our thinking may be the reuse or
simulation of our experiences in the environment [..]
Representations allow us to re-present things to ourselves
in the absence of the thing, so that we can think about it, not just
react to it."
Intelligent behavior is characterized by prediction and
imagination, intentional action and reasoning. Davis continues:
"Animal intelligence has a here and now character: With animal calls,
for example, there is an immediate link from the perception to the mind
state to the action. If a monkey sees a leopard, a certain mind state
ensues, and a certain behavior (giving the appropriate call) immediately
follows. Human thought, by contrast, has an unlimited spatiotemporal
reference, by virtue of several important disconnections. Human thought
involves the ability to imagine, the ability to think about something in the
absence of perceptual input, and the ability to imagine without reacting."
Therefore building intelligent agents and creating a virtual "Matrix" world
are in fact closely related topics. Today there are sophisticated 3D
graphics engines, especially in new computer games (for example
Far Cry), which can display very complex worlds. Thus the goal
of AI seems so close as never before. James F. Allen has written
more than 5 years ago in the AI Magazine article "AI Growing Up -
The Changes and Opportunities" (Volume 19 Number 4 (1998) 13-23)
"We are [in AI] at a similar transition point to the first flight in
aviation. The field of aviation was changed dramatically by the
development of working prototypes because for the first time,
experimental work could be supported [..] I believe that we're at
a similar transition point to the first flight because we are now
able to construct simple working artifacts which then can be
used to support experimental work."
This experimental work should answer unresolved questions,
for example the question Nils J. Nilsson formulated in "Eye on the Prize":
"Is general intelligence dependent on just a few weak methods
(some still to be discovered) plus lots and lots of commonsense
knowledge? Does it depend on perhaps hundreds or thousands
of specialized minicompetences in a heterarchical society of
mind? No one knows the answers to questions such as these,
and only experiments and trials will provide these answers."
Allen writes further in his article, that a critical point in the
transition is the development of a calculus or a set of laws:
"By analogy to a mature science such as physics, we are [in AI] at
a stage prior to the development of calculus and Newton's laws."
What calculus could that be ? I am convinced it should
be based on the language of Multi-Agent Systems (see
Minsky's classic book "the Society of Mind"), both for
the external and the internal world of the agent. Distributed Artificial
Intelligence (DAI) has become the most important branch of AI.
Probably Allen is right, and we are at a critical transition point,
similar to the first plane and the first flight in aviation. The computers,
graphics engines, machines and cluster seem to be strong enough.
Yet I have the impression, there are in fact many people working
in AI, but too many people who want to construct the cockpit
(especially the console of the cockpit), and each of them tries to
create it alone, whereas there are not enough scientists who work
_together_ on the engine and the wings, for example in a common
large scale project.
The engine and the wings correspond to the software which enables
the agent to move around in a complex 3D world and which permits
the agent to understand this complicated world (including a powerful
and modular 3D-graphics engine). Such a distributed application
should at least enable the ability to understand the prepositions
at/on/in, before/behind, above/below, over/under, etc. (*)
Allen's attempt of a one-sentence AI definition is "AI is the science of
making machines do tasks that humans can do or try to do."
Humans live in a complex multi dimensional environment, and the main
purpose of the brain is to control the movement of the body in this
environment. If we are able to construct agents with common sense
which are able to understand a complex 3D environment, we come close
to AI's original goal, which is to produce intelligent programs that are
able to
use general tools and to build systems with humanlike capabilities and
intelligence.
An intelligent agent with common sense should have at least the
_common_ abilities of all humans. It should be able to do what every
human can: everyone of us is an expert in speech & language and
vision & motion, with the highly developed ability to understand and
comprehend a complex 3D environment.
(*) see
Language and spatial cognition
Annette Herskovits
Cambridge University Press, 1986
[ comp.ai is moderated. To submit, just post and be patient, or if ]
[ that fails mail your article to , and ]
[ ask your news administrator to fix the problems with your system. ]
--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)
|
