AGENTS and INTELLIGENCE
... and Definition Making
by Adam Maria Gadomski
White e-Paper, since 1997- 2001


 

  "...one of the central goals of AI is to develop artificial 
   agents that embody all the components of intelligence ..."

                             [ J.Laird & P.S.Rosenbloom, AI Mag.,Winter 1991] 

 

The synthesis of available on the Web definitions according to the TOGA meta-theory

(Top-down Object-based Goal-oriented Approach)

   We may start from the following top intuitive definitions of intelligent systems:

  Intelligent Systems are systems recognized arbitrarily by humans, as "intelligent". They, in an integrated manner, exhibit mental, human like, reasoning functions and act with certain efficacy.

For the reason of this "certain" efficacy, this term is used arbitrarily in biology, cognitive and social sciences, and by computer scientists, software developers and business people.

  In computational sciences, the intelligent system kernel/mind/brain functioning is frequently modeled  as a multi-agent system (MAS) or an abstract intelligent agent  (AIA), i.e. abstracted from its possible physical realizations, individual experience and environmental constrains.

  Artificial Intelligent Systems are goal-oriented/driven/directed/based or purposive, autonomous agents (in the range of their competences and obtained tasks), and are realized in available AI and software technologies.

 

  In general, in computer science, already the concept intelligence  is ill defined, see: cognitive intelligence, and:

"...These expert systems (whether neural nets or not) are truly machines in the old-fashioned sense. They are housed in specific hunks of metal and silicon, and fed carefully-culled information in meticulously and specifically prepared chunks. These exhibit "intelligence" in only a weak way, regurgitating intelligence, like a child imitating an adult. This lack of meaningful intelligence is patent when programs randomly combine words and then edit those according to rules of programmed grammar to generate "poetry." The same lack of meaningful intelligence is evident in programs which, given the rules of logic, spit out syllogisms, or make a correlation between objects whose common properties were already programmed." 
[Curtis E.A. Karnow, 1996  http://www.law.berkeley.edu/journals/btlj/articles/vol11/Karnow.pdf].

  Agents are systems, and some agents can be considered intelligent.

   A synthetic, most general functional definition of an agent according to the TOGA  systemics human-oriented perspective:
 
 Agent is a self-active system, i.e. it always works (==always active) for a monitoring (observation) and interaction with its own dynamic environment according to the externally predefined tasks/goals/preferences For example, an agent may realise one or more routine functions of an agent system.
The manner, domain and range of th
e monitoring function, and  agent interactions with its environment (i.e. its behavior) can be differ and variable with time.  Contrary to the properties of physical bodies, agent's reactions depend on its internal dynamic state (i.e. its modifiable memory) and not only on the state of its environment. 

  Agents recognize some attributes of their environment and act according to their individual preferences, competencies (knowledge) and available information .

 

  Remarks: 

  - "Agent" is the name of an abstract concept and its definition depends on the consensus of researchers and practitioners which  

     use it. Therefore, in different contexts it has many  not functional but structural definitions expressed in local terminologies/ontologies.

  - The clear formal distinguishing  the concepts: information, preferences, knowledge (*) is  essential for the domain- independent

    definition of an agent in the TOGA meta-theory . This assumption is also explained and motivated in the EU Sophocles Project

    Report [Adam Maria Gadomski,2002].

  ---------------------

  (*) Of course, the names of these three concepts can be different, for example, somebody, can call them: (property A, property B, Property C), or using local  metaphors: (data, wisdom, knowledge),  ( information, motivation, experience), and others.

 

  From  Abstract Intelligent Agent: PARADIGMS, FOUNDATIONS AND CONCEPTUALIZATION PROBLEMS (1994):

     "An abstract intelligent agent (AIA) is a hypothetical model that captures the essence of all systems which we accept as "intelligent" in the  common sense. We review different perspectives on agents discussed at the Round -Table <AIA1993>, including the difference between agents and physical objects, agent behavior, observation and action capacities, communication with other agents, architecture and reasoning. We define the concepts of agent, abstraction, autonomy and intelligence, and we discuss how the behavioral, social, and functional context contribute to these notions. We envision AIA as an algorithm which can be shared among many carriers and which is unspecific about  concrete links with the real world. We confront several concepts of autonomy, which emphasize the means of the agent, agent's self-reliance, and capability of acting independently from other agents. We analyze different notions of intelligence and their compatibility with the notion of a universal abstract intelligent agent. Finally we summarize the AIA research program which calls for an abstract dynamic architecture capturing the foundation of intelligence, and for the specification of physical systems which can be  carriers and activators of this structure." [Gadomski, Zytkow.1993]

  See  also:  Yahoo: AIA Discussion Group.

 

  Agency,  this term is frequently used in the subject matter literature. It usually denotes a collection of agents. According to the TOGA   assumptions we may accept the following functional definition of agency:
 

 Agency is a system/organization of agents which all together may exhibit  all necessary properties of an abstract agent. 

It means, not all organizations of agents  are agencies.
 
  Some external links

 - Abstract Intelligent Agent: Yahoo Discussion Group (since 1998)
  - Intelligent Agent Architecture. Entry by Stanley J. Rosenschein in the MIT Encyclopedia
  - Cognitive Inteligence

  - CALT Encyclopedia - agent technology: www.insead.fr/CALT/Encyclopedia/ComputerSciences/Agents/
  -
Google search:  cognitive, intelligence, agent 

                               "Intelligent Agent Technology" - Worldwide IEEE / WIC / ACM International Conferences.

 

   See more on this server

 

  (The concept of Agent and Intelligent Agent (1997-8)

Personoids - Intelligent IPK Agents

 Intelligent Agent Ethics

MKEM Server Home 

 

© 1997-2007, ENEA. A.M.Gadomski. All rights reserved. No permission is granted to download and save professional images, code, or any other material from this server other than for viewing and citation purposes.  These are research pages representing the opinions of the contributors, but not necessarily of ENEA.