Generally, Artificial General Intelligence is considered as an artificial intelligence configuration that is capable of operating at the sa...
Generally, Artificial General Intelligence is considered as an artificial intelligence configuration that is capable of operating at the same level as human cognitive abilities. This approach sets the goal of Artificial General Intelligence to realize universal thinking abilities, as human-like cognitive abilities are considered a form of universal intelligence. Humans are capable of abstract thinking, which definitely enables universality, but considering only human-level intelligence as general intelligence would be too restrictive of the concept of generality.
General intelligence is an ability that humans are capable of at some level. To achieve Artificial General Intelligence, human intelligence can be used as an example and a model, but it is too limited to be considered as a goal.
In the search for a way to achieve Artificial General Intelligence, it is necessary to define fundamental principles about cognitive abilities that can be applied to human intelligence, but are more general, not limited only to actual human abilities. General intelligence is a more general concept than the realization of human capabilities.
What are the basic interpretations of principles relevant to general intelligence?
Intelligence: problem-solving thinking.
Thinking: searching for relations between states.
State: configuration, ordered structure, arranged order.
Problem: a goal of finding relationships between particular states.
There is not necessarily a relationship between specified states, therefore not every problem can be solved. Furthermore, the relationship between states can be direct or indirect, i.e. through the interrelation of other states.
States can be represented by their characteristic description, by the information that is specific to them. The ideal form of information is a state in a format that can be universally applied to represent any state. One such ideal information representation is the number format, whose consistent (non-contradictory) manipulation is called mathematics. When mathematics applies the rules of how the real world works, the consistent manipulation of information in the number format conspicuously represents the states of the world and the changes of states of the world with precise accuracy. Computers use information in number format to represent states.
Similarly, human language can be viewed as a more or less ideal informational representation of states. Human language is built on concepts based on generalizations of meaning derived from the functioning of the human brain, and represents states and changes of states by connecting concepts and consistently manipulating their relationships. If human language takes into account the laws of the real world, then human language, as a universal information format, is suitable for representing states in the world, and furthermore, the consistent application of human language is also suitable for properly representing changes in the world. Humans use information in the form of language to represent states.
General intelligence: an intelligent thinking ability capable of recognizing the correlations (co-occurrences) between states and the hierarchies between correlations (the relative arrangement of correlations, the connections between correlations relative to each other) using a universally applicable information representation method capable of describing the states, and which is capable of applying the gained knowledge acquired by recognizing the correlations and relations to solve problems, capable of behaving intelligently to achieve defined states designated by an external intention, i.e., capable of defining the transitions between the current state and the designated state through its acquired knowledge by applying a consistent transformation of states in accordance with the rules of the learned system in the format of the universal information representation used for describing the states.
The level of general intelligence is represented by the difference in the ability to recognize correlations and relations between information representations according to the distance of the relations, i.e. in the depth of the connections.
Human intelligence is a form of general intelligence of a certain level of development. The intelligence of human society is an even more advanced form of general intelligence. A computational system that uses numbers as a universal representation of information can, through the application of appropriate mathematical procedures, have potentially limitless general intelligence capabilities.
Human intelligence is based on the functioning of the brain. The functioning of the human brain effectively enables the carrying of intelligent abilities. However, the human brain does not operate according to the operating principle defined by the arithmetic rules of the binary number system, even though there are similarities in the operation of the classical computer and the natural brain. A data processing system using procedures based on the operating principle of the brain could increase the operational efficiency of an architecture capable of creating general intelligence. Given the striking similarity between the existence of the quantum world and the natural functioning of the brain, both based on resonance, the use of quantum mechanics to create artificial general intelligence using brain-like natural operating mechanisms could allow us to achieve the most efficient intelligence configuration and the most effective general intelligence possible in our physical world.
No comments