International Encyclopedia of Systems and Cybernetics

To put this very general definition by ÖREN in perspective, let us quote R. FISCHER: "The most important is that the brain is self-determined, and cannot be programmed, without the consent of its Supervisor-Ego" (1992).

Leaving aside this last conceptual crutch, the self-determining brain can be understood as self-organizing, in terms of MATURANA, VARELA or VENDRYES' autonomy. Conversely, artificial intelligence in its classic sense, is not self-organizing. This could however not be anymore true for future connection machines, constructed as parallel working networks.

ÖREN, having thus defined intelligence in very general terms, elaborates in the following way:

"… an advanced knowledge-processing ability implies the intelligence of an entity to be an ability:

1) for the representation of knowledge about itself and/or about some other entities;

2) for the generation, transformation, organization, location, selection, evaluation and analysis of different types of knowledge, as well as

3) for communication. The ability for communication requires the ability to answer and ask questions in a language (computer or natural language, including a spoken language) and includes abilities for perception, sensation, elicitation, acquisition, assimilation and issemination of knowledge.

"Being an adaptative knowledge processing ability, implies that the intelligence of an entity is an ability for changing its environment or itself. The latter implies that intelligence is an ability:

1) for learning, which in turn implies abilities for generalization, abstraction and specialization of knowledge, and

2) for mutation, i.e. generation of self-modified versions of itself.

"Being a goal-directed knowledge-processing ability implies the intelligence to be an ability for:

1) goal setting

2) goal processing (including subgoaling, goal modification, goal ranking, goal selection), and

3) goal-directed knowledge-processing (including goal-directed representation, generation, transformation, organization, location, selection, evaluation, and analysis of knowledge as well as communication and adaptation).

"A definition of artificial intelligence can then be: "The science and technology of knowledge processing using software that has goal-directed, adaptative, and other advanced knowledge-processing abilities. The aim of artificial intelligence can then be:

1) to find out about the nature of device-independent intelligence, and

2) to build intelligent machines (i.e. intelligent computers and intelligent computer-embedded machines) via advanced software" (p.5/6).

There remain, of course, lots of question-marks; for example:

- What is precisely knowledge, compared for example with stored data, information, or wisdom?

- How are goals formed? If a machine becomes able to form goals, what would be the interrelations between natural and artificial intelligent beings?

- To which extent is intelligence (natural or artificial) algorithmic after the necessary learning process?

- What does precisely self-modification mean?