So What Is An Agent?

The term Agent is used by many people to mean many things for many different reasons. This section will, in time, contain alist of agent definitions and an analysis of these definitions: I (somewhat pretentiously) call it An Ontological Specification of the Concept of Agency . AI is a toolkit of techniques of use in solving computer or digital information problems. Computational Intelligence covers a wider spectra of activities. Cognitive Intelligence ,typically, draws its inspiration from non-synthetic intelligence.

An integrated computational entity with intentionality and some degree of autonomy: [Davis; 1996; ECAI-96]
An agent is a synthetic entity that enables us to study, at a computational, design or theoretical level, what a mind could and can be.: [Davis; 1998: IEEE SMC-98]
An entity is a software agent if and only if it communicates correctly with its peers by exchanging messages in an agent communication language”,: [Genesereth & Ketchpel; 1994; Communications of the ACM, 37(7), 48-53]
Autonomous agents are systems capable of autonomous, purposeful action in the real world: [Brustolini; 1991; Carnegie-Mellon University, Technical report, CMU-CS-91-204]
An agent is anything that can be viewed as perceiving its environment through sensors and acting upon that environment through effectors: Russell & Norvig; 1995; Artificial Intelligence:A Modern Approach]

Integrated Computational Entity: An agent is holistic computational process, but can be distributed or societal. At some level of analysis, a group of distributed processes can be recognised as forming some entity.
Intentionality: This enables a correspondence between internal and external states, and requires agents to perform purposeful action
Autonomy: Agents operate without direct interaction of other entities and have (some form of) control over their actions and internal states.
Reactivity: Agents perceive their environment and respond in an expedient manner to changes not necessary for all agents
Temporal Continuity: This is really emergent property of the above four attributes.
Flexibility: Actions are not scripted or predetermined

Reasoning about Beliefs: Inspite of (my) many attempts to bury logic and go for lunch, here is a reference to formal (logic) models relating belief and logic in agents.
Communication: Agents can communicate via some agent communication language
Social Ability: Agents cooperate or compete towards collective goals (Probably necessary for multi-agent societies)
Adaptability: The actions of an agent are not scripted or predetermined and they can modify their behaviour
Learning: The capabilities and behaviours of agents are not static. They can modify their (internal and external) actions through experience. This is a more expansive term than adaptability or flexibility.
Pro-activity: Agents do not simply respond to events but take initiatives. For example, they identify possible problems before occur in an attempt to preclude undesired events occuring.
Self Actuation: Agents (attempt to) define their own agenda and redefine their position within their environment or a new or expnanded environment. Consciousness; Oh Brother - is this one troublesome! See Richard Rorty : Philosophy and The Mirror of Nature - You are nothing but a set of intermeshing biological weaves somehat precariously controlled (?!?) by (mis-)behaving neuronal complexes. A more serious answer is not available from here.

[Picture]