Intelligent Agents......

[Alain Farmer]

Hello Pierre Sahores, Chipp Walters, and
y'all who are interested in MetaCard-based AGENTS

>> w/out going too much into it, can you explain
>> what an Intelligent Agent does in MC? Perhaps
>> an example? I'm interested in learning more:-)

When discussing software agents, as we are now, one
has to almost immediately talk about agent
*architectures*, because software agents, unlike their
human counterparts in 'espionnage' circles, are
especially interesting when they work together as
coordinated teams.

I will attempt, nonetheless, to define a single
"agent". It's a software program that can autonomously
accomplish the task it was crafted-or-evolved for. It
automatically adapts to unexpected changes in its
working environment (the Internet, an intranet, or a
single desktop system). 
Because these changes are unexpected, a rigid
algorithm does not make the grade. Unexpected
situations & changes require an "intelligent"
adaptation to maximize gains & minimize losses. And
this is all happening dynamically, of course, which
means that the agent has to be active and perpetually
reviewing itself ( its programming and the agent's
impacts on its environment, including other agents ).

In sum, here are the capabilities an agent needs
to be truly autonomous. The agent must :

* be conscious of its environment ;
* be aware of its self ;
* be able to grow and prune its knowledge ;
* be able to transform itself (adaptation) ;
* be able to transform itself over time (evolution) ;
* coordinate its activities with other agents ;
* communicate fluently with human-user[s] ;
* adapt to the human-users' preferences, habits, etc.

Hummm ... Pretty good definition of a single agent
after all, albeit I alluded to "other agents" several
times in my above description.

The main reason why having MORE THAN ONE AGENT working
together is a **really** good idea is that human-level
intelligence is [nearly] impossible to achieve as
single programs. According to Marvin Minsky in the
"Society of Mind", and other authors too, even *human*
intelligence may in fact be a variety of different
intelligences that work together synergetically to
form 'The One' that we are aware of.

The theoretical foundations of the above rest upon
some very solid principles of hard as well as 'soft'
science. Namely: the dynamics of open-ended systems
that are far from equilibrium. To deal with a constant
influx of new energy, these self-organizing systems
evolve dissipative structures (order) that *reverse*
the universal tendency towards entropy (disorder).

The gist of it is that *order* emerges [by chance]
from chaos. This spontaneous order feeds on itself to
create ever higher forms of order. Collective
properties emerge at a systemic level that are NOT
sustained by any given component but is more like a
regularity due to the high number of *interactions
between* the components. As the old saying goes : "The
whole is greater than the sum of its parts".

On a practical basis, this means that we can create AI
systems by creating societies of coordinated
autonomous agents. Each singular agent can/should be a
very simple program that accomplishes one very modest
goal. Complex goals are accomplished by making the
agents collaborate amongst themselves. Not a rigid
architecture, where all the links are PRE-established,
but rather a runtime env where the relationships among
the agents are negotiated amongst themselves (and
guided by the user, of course). IOW, a dynamic
open-ended society of agents from which systemic
properties will emerge from their interactions with
themselves and their environment.

If you're thinking that this is how *human*
intelligence came about, in our own [social]
evolution, then you are indeed an astute
reader/thinker. Btw, YES, I am indeed saying that
intelligence is a emergent systemic property AND,
furthermore, I suspect that [our] intelligence is a
*collective* property, e.g. shared by many
'components'  at once (humans in this case) but not in
any particular component of the system! R-a-d-i-c-a-l,
eh!  ;-)

IOW : If it were not for y'all, I'd be pretty dumb! As
would anyone else be in a singular world. We mutually
define and expand each others' minds. It just occurred
to me that this is a bit like the matrix. You have
just been told that you do not have a singular
intelligence, albeit you always thought you had, up
til now that is! You have selected the blue pill
instead of the red one. You cannot turn back, Alice.
The matrix (society) *is* your mind. Well .. not
"yours" actually .. everyones!

Life is such a TRIP!  :))

Alain

__________________________________
Do you Yahoo!?
Yahoo! Finance: Get your refund fast by filing online.
http://taxes.yahoo.com/filing.html