What is True Intelligence?

By: Mahmood Shamizi
Introducing the writer
The author of this article has written some more articles and short essays
on Artificial Intelligence (AI) and throughout all these essays the
possibility of creating an artificial true intelligence is discussed which later
leads to a new discovery on the nature and the essence of Intelligence.
The best article written by this writer related to the same topic was titled
“Non-Mechanized Vision”. This was presented in an International
Conference called Machine Vision Image Processing and Applications,
M.V.I. P2001 held in Birjand from 7- 9 March 2001, which was very well
received by the renowned professors in the same discipline. A letter of
appreciation was later written and given to me by the secretary of the
Conference. In the introductory session, a sample robot was demonstrated
in the work shop of the conference that was working on the very same
principles and was able to interpret / analyze the input data.
What is the True Intelligence?
My goal in writing this article is to present a concise and accurate
definition on Intelligence; either artificial or natural. This definition will
be henceforth referred to as ‘true intelligence’ to imply the qualitative
difference between ‘true intelligence’ and machine-like actions.
Should we fail to define a precise border line between the meanings of true
Intelligence and mechanized actions and define the distinct/clear
characteristics of both in order to differentiate them from one another, we
cannot be optimistic on the continuation of scientific advances in AI.
One of the major factors that has prevented us from doing proper research
on the true nature of Intelligence, originates, in my opinion, from the fact
that Artificial Intelligence experts are not so keen on analyzing the true
meaning of Intelligence and that they prefer to concentrate on studying
non-intelligent systems, control systems, and their pertinent governing
laws. This, I believe, is most certainly a diversion from the main idea and
insufficient to define Intelligence, for in non-intelligent systems the
capability of creating a sophisticated behavior does not exist and that every
behavior has already been thought of and preset in the system as an inbuilt
property into the system by the designer. But a true intelligent
system should, itself, be a creator and developer of new
behaviors. Therefore, the governing laws in such systems will have
qualitative differences with other systems. Thus, the expert must have
studied the important issue of the mechanism of creation, the way which is
demonstrated, and the origin of emergence of intelligent deeds
To take a true intelligent system equivalent to a mocking system on one
hand and the efforts made by some experts to simulate intelligent
behaviors by robots on the other, has made the study of ‘true intelligence’
less appealing. Therefore, scientists are perhaps losing interest in the
realistic meaning of the true Intelligence. It should be added here that
obviously, seeing similar behaviors from two systems does not imply that
their internal construction are also the same. The two mechanisms may
look alike, but they surely have major qualitative differences.
In other words, it is possible to see an intelligent act rendered by a nonintelligent
source and actually achieve the desired goal, but we may also
see an intelligent system that may experience an undesirable result from a
wrong action. It is therefore concluded that one must not conclude that a
system is intelligent simply because it has rendered a correct action or,
otherwise, non-intelligent simply because it has made a false move. One
must look into the mechanism of the original selector and ascertain if the
system is actually aware of the reason why a particular action was taken or
otherwise. If not, the action has been a hap hazard or machine-like one and
that it has been made blindly or automatically. Because the system itself
does not know why it has executed a particular action and is not even
conscious if the act has taken place, for it has done a compulsory task.
But if a system actually knows why it has decided and rendered a particular
job by its own reasoning, it can be said that the job has been done wisely
and out of its own free will. Therefore, it is justified to say that the action
has been carried out deliberately and intelligently.
To prove this statement, two reasons can be given;
1. If for a sophisticated question a wise answer were already
installed in a computer, it is only obvious that the given wise answer
is not originated out of the computer’s awareness, intelligence, talent
and comprehension. Moreover, the replying mechanism is not
comparable to that of a scientist. Because, the relationship between
the question and answer is not at all based on the knowledge and
intelligence, but it is based on the designer’s choice who has already
preset the answer within the system. Such systems are not aware or
conscious as to why they have given the right answer or have made
the right move. If we let a computer be free to chose the right
answer amongst few optional answers in order to find the true
answer by trial and error method, we can justifiably assume that the
choice of an answer will not be made according to the meaning of the
text and/or the content of the answer, because the answer(s) are
already preset in the system by the designer.
If some people still persist that intelligence has similar features to
above, the examples of doors that are automatically opened when
approached by a person as well as the reaction of a thermostat to
heat must also be counted as the intelligence of the automatic door
or the reaction of thermostat. The expansion of materials as a result
of temperature rise can also be argued as being an intelligent act…
In this case, a great many physical and chemical phenomena may be
interpreted as some type of intelligence and thus devaluate the
concept of the intelligence down to include simple substances like,
metals, minerals and so on.
Where has this degradation and trivialization of the concept of
intelligence originated from? This originates from the fact that we
have not been able to define a distinct border line between the
concept of the truly intelligent systems and unintelligent systems
and provide a qualitative characteristic for differentiating the true
intelligence from the systems that render machine-like, reactionary
and blind manifestations.
2. The present approach to the meaning of Artificial Intelligence in
the scientific societies focuses more on its action oriented aspect
nowadays rather than its operation oriented aspect, whence it is
known that operational aspect is not one of the fundamental
conditions for an intelligent entity.
Let us visualize an intelligent entity that is busy observing a far off
environment to analyze and understand all that is going on there,
without intervening in that environment in order to determine the
prevailed governing laws. This is Just like a human being who is
watching heavenly bodies through a telescope to discover the
prevailed governing relations without actually intervening in the
environment.
To define intelligence, we must not, therefore, put too much stress
on its operational-orientation aspect, despite the fact that having
operational ability helps the system to examine its own ideas which
will lead it to acquire better and more precise know-how of its own
environment.
Therefore, observing intelligent and correct behaviors from a system
does not substantiate the system’s intelligence and one must look
into its internal structure and its mechanism of selecting a course of
an action and whether it is originated out of the system’s awareness
or otherwise. This awareness is only possible if the system were
able to analyze the acquired information received from the
environment by itself and can find meaningful relationships between
the cause and the effect within the environmental accidents. In other
words, the system should be able to create models, no matter how
incorrect, on the basis of its own observations from the
environment. Then the system should create a valuable idea from its
own impression and launch it as a desired target. It will then look
for finding and linking appropriate tools to reach that goal and
thereby create an operational algorithm to fulfill that whim.
Please note one of the major characteristic of an intelligent system is:
creation of a goal before devising an operational algorithm.







We can now present a definition for intelligence that: ‘whenever a system
discovers an order in the data that it receives and thus becomes able to
detect conditions that have applications in guiding the system to its goals
and values, it can be said that system has virtually given meanings to the
data or has explained the data for itself;”
True Intelligence”.
Therefore, it has found new connections between the incoming input
data and an internal value and that it has interpreted the incoming
input data into a bunch of useful information for its own self. This
capability of the system in acquiring actual know-how is equivalent
to constructing a mental model from its environment. Obviously,
full compatibility of a mental model with the current truth in the
environment is not imperative but having such capability is a reason
for the existence of true intelligence in that system.
Whenever a designer anticipates certain reactions of the system with
respect to different inputs, no more ideas or even no new algorithm
will be developed or created its realization. Consequently, all the
system’s reactions will be compulsive, machine-like and blindfolded.
We can only call a system intelligent that is free to take any
actions and interpret the input data freely, meaning that
recognition is only made by the system itself and that the selection of
the course of actions is made based on the impression and
understanding of the system from its environment. Further, the
usefulness of the actions is judged according to the system’s selection.
On the contrary, if we force a system to follow the programs and
algorithms installed in it, the liberty of action would be obscure &
meaningless. When a system is not free in making decisions and
selecting its own course of actions, how else can it manifest its power
of creativity, talent, inference, and intelligence?
It may seem impossible to construct a system like above, due to the
fact that there is some kind of antagonism in above remarks and it
may seem that we are trying to write an algorithm for a system that
is not obligatory for it to abide or to obey. Alternatively, it may seem
that we are developing a system without an algorithm.
But with a little care we’ll see a delicate and mid- way solution to
escape from the above paradox, and that is to say that we want to
write an algorithm that is able to create and construct other types of
algorithms itself by forming a mixture of the system’s observations
and past experiences. This algorithm would be the decisive factor of
the deeds and behaviors of the system, of which the designer would
have no more knowledge. For this very reason, the system’s
reactions cannot be anticipated. Although the evolved algorithm
may not be a perfect one in the beginning, the system will lead itself
towards sophistication provided that the road of improvement and
dialectic evolution were left open. From then onward, the system
can create logical cells and thereby succeed in making mental models
according to its observations from the environment.
Only a system like above would be capable of producing ingenious
solutions and new ideas for problems with the help of its own
abilities and the knowledge that it had acquired. One must not
forget this important issue that for an intelligent system as such, the
old and obvious ideas of ours are all new discoveries for the system,
because it has discovered and understood them on its own and that it
has created them afresh by itself. This ability of creating ideas
liberates us from installing logics or the governing laws of
environment into the system, and if the system were placed in an odd
and different environment, it will be well able to extract the
governing laws of that new environment on its own.
Expansion of created algorithms in the system will be of two types:
First Type: Cognitive:
Expansion of the created algorithms to describe the input data
and to upgrade the ability of producing links between the goals and
the received input data and how this input data are interpreted and
the processing method of this input data are processed of explaining
the input data are analyzed. In fact, interpreting the input data and
comprehending the location and the atmosphere in which the system
is situated is the reason of constructing a mental model from the
environment and more perfect ideas.
Type 2, Behaviorist:
Algorithm expansion in order to implement ideas in the environment
through gaining experience in operational steps means: “the ability
of creating links between the tools and requirements or
ideas”. Thus, if a system acts on the basis of its observations,
impressions, and inferences, then it has the freedom of will, that is to
say, its reactions cannot be speculated just by its original in-built
algorithms. Similarly, it is not possible to speculate how the input
data were processed. In other words, the exact speculation of its
reactions is not possible due to the following reasons:
a) Designer is not aware of the system’s observations throughout
its life span.
b) Designer does not know the hierarchy, and the type of the
observations made, and the incidents that the system has been
encountered with.
c) Designer does not know how the system has analyzed its past
observations
d) Designer is not familiar with the governing laws in the
environment wherein the system is situated.
e) Designer is not familiar with the new values that are
permanently produced based on the mental functions, reasoning,
and studies made by the system.
f) Inability of the system in grasping precise understanding with
absolute truth as well as flawless impressions from the current
logic in the environment will cause some erroneous interpretation
in explaining the governing environmental relations. Therefore,
it is certain that some differences will exist between the methods
of reasoning and inferences of the designer and that of the system,
which implies freedom of reasoning and thoughts of the
entity. In other words, the types of the acquired impressions and
erroneous actions of two intelligent systems cannot possibly be
the same everywhere.
(1)
The overall conclusion is that one can not rely on mathematical models to
build such a system. On the other hand, the current neural networks do
not serve our purpose either, since their reactions are predictable and that
they are bound to comply with the certain mathematical rules. Besides, the
known logical mechanism of theirs do not allow a system be created with
the freedom of will, but it is possible to develop a backbone that would give
birth to any kind of algorithm that is necessary and let its cells to undergo
a dialectic growth and thereby develop large textures that are open to any
development for the system to meet its appropriate demands. Due to the
uncertainty of its internal logics
(2)
It is therefore seen that to construct an intelligent system, mathematical
models can not be relied on. Then, the currently used neural networks
may not be deployed because neural network’s reactions are
predictable. Thus, if a system were obliged to comply with mathematical
and predictable laws, while being aware of the logical performance
mechanism, it will no longer be a system with freedom of action. But we
can write a basic algorithm to form a backbone for production of any
algorithm thereafter for dialectic growth of its cells and large textures,
such that to keep the evolution path open for creating various assemblies
that are appropriate for the system’s needs.
Due to uncertainty of the internal logics and non-dogmatic mental rules in
some of the beliefs, the system hesitates and is unsure on the outcomes of
its decisions.
This hesitancy is another reason why the evolution path is being left open,
and a reason for freedom of choice to select any possible rout. In fact, only
a system like this has the freedom of will and thought.