|Intelligent Systems And Their Societies||Walter Fritz|
In our review of intelligent systems and their societies, we started with tools. We indicated why tools are so important to our standard of living and how the different types of tools we have invented, have increased our standard of living enormously. The last step in this long process is a certain type of tool, namely machines that think.
But what do we mean by thinking? Not everyone has the same concept for a given word. We must say what we mean when we use a certain word; otherwise we will be misunderstood. So we highlighted the importance of having precisely defined concepts.
In this site, we have talked continuously about systems. But what are systems? So we defined systems as being a part of the universe, containing parts that have stronger correlations with each other than with the rest of the universe. While looking for this definition, we found that there are really two types of causes. One type is related to the changes caused by an intelligent system, and the other type is related to those changes occurring in nature, where one occurrence only has a correlation with another.
In studying "intelligence" we have not limited ourselves to one aspect, such as vision, problem solving, or expert systems, but we have studied the total intelligent system. This system has senses, objectives, a selection of responses from memory, a possibility to act on its environment, and finally the ability to learn from its experiences. Thus intelligence, in the sense that we defined it, is basically a stimulus - response mechanism, but with a selection of responses according to an objective. It is a way to chose an adequate response to a given situation, a response that brings the system nearer to its objective.
This study has given us a definition of the intelligent system (IS), applicable both to artificial and natural intelligent systems. An IS is a system that learns during its existence how to reach its objectives and in doing so, consumes energy.
It is amazing how little an IS, both natural and artificial, can really know about its environment. The IS builds up an internal representation of its environment to the best of its ability and in doing so, creates its own concepts. Surprisingly, we have seen that in both the natural and the artificial IS, concepts are something material, not immaterial or eternal entities.
In both the natural and the artificial IS, it is of the utmost importance that incoming information be checked. We discovered that while sensory information is often limited, it is (mostly) reliable. Information received from another IS, however, is often incorrect, and most often unintentionally so. It must thus be carefully checked.
Therefore, three aspects of knowledge exist:
We examined in some detail the interior workings of an artificial IS and the many similarities to natural ISs. We saw how an artificial IS creates concepts and stores its experiences as response rules. We saw how the IS compares the present situation with its stored response rules and how it chooses one rule and then performs the response part of that response rule. Finally, it generalizes the response rules by vertical and horizontal comparisons, to make them applicable to many different situations.
The IS interconnects concepts creating a web. Concepts have links to those related concepts that are more abstract, and to those that are more concrete. Also each concept has a link to its parts and to those concepts it is a part of.
It is amazing that in both natural and artificial ISs, a kind of stimulus - response mechanism, used to reach an objective, is sufficient to explain almost all of the occurrences in the brain. Even though it uses a complex and sophisticated mechanism, still, it seems just a special type of stimulus - response arc.
We have examined a theory of mental processes (of thought), experimentally, in our artificial ISs. We have seen some of the many artificial ISs, written by other people. We noticed many fundamental similarities and some differences in detail.
It was interesting to note the many similarities between artificial and natural IS, and to learn of their differences. In today's artificial ISs, emotions are not used or used only sparingly. Surprisingly, in both humans and animals, emotions play an important role as an indicator of whether an objective was reached or not.
Human beings are slaves of their habits. Some habits are very general response rules that govern all of our behavior and thus give a person a "character" or a "mindscape".
Studying the IS, we saw that such a system is aware of some response rules and concepts, namely those applicable to the present situation. However, it is not aware, in a given moment, of all others. Also, the basic mechanisms of handling concepts and response rules are unconscious.
We have, not quite arbitrarily, defined a society as a system that has objectives, and is composed of ISs having an appreciably shorter life span than the society. This definition fills the same role as the arbitrary definition of "triangle" does for geometry. It gives us the tools for a mathematical treatment of the occurrences in a society.
We saw the role that the governing sub society plays within a society, and that the usefulness of the division of labor is due to the limited learning time of ISs.
We proposed a political subdivision of any society into a hierarchy of parts, where each part has a similarity of function with the next higher whole. This provides a great advantage, since it permits the training of governors at each level of the society. We also saw that the only effective way to avoid war is by making each nation part of a higher level of society, just as today's provinces are part of a nation.
Closely related to a society is the knowledge of ethics that can, having defined the IS, be transformed from a religious or philosophical teaching to a science. This conclusion leads us to say: "Insofar as a human being is an IS it should do, in this situation, what we calculate that an IS would do". Any other action, by definition, will be worse for the attainment of its objectives. The controversy between determinism and free will in human actions is now resolved. We saw how each choice of response is limited to those response rules that exist in an IS and why the choice has, necessarily, an element of chance.
Sociology and ethics can now be developed as hard sciences using numerical values, instead of using only qualitative observations.
The study of artificial ISs has also shown what their "world view" could be. Again we see that what is true for the artificial IS is also true for human beings. This conclusion changes philosophy into a hard science.
We saw that it is unreasonable to fear artificial ISs. They are just machines, designed by us to serve us. Even when they eventually become more intelligent than we are, they will not be dangerous since we are designing them to be our servants (Their objective "to serve us" is built in).
We human beings are starting another industrial revolution, with a drastic reduction in working hours and a drastic increase in our standard of living. We have a solution for coping with these changes, namely the social dividend.
Finally, it is useful to learn from the artificial IS and thereby increase our own intelligence.
It is amazing to observe how a small thing such as a rigorous definition of an intelligent system can have ramifications on our understanding of societies, ethics, and philosophical outlook.
For continuous reading, like a book - continue here.
Jump to the e-book Contents / top of this page.
Copyright © New Horizons Press