Download intelligence and intelligent machines

INTELLIGENCE AND INTELLIGENT MACHINES: A VIRTUAL
REALITY?
by
Winfried K. Rudloff
Governors State University
Keynote Address Presented at InterSymp’94
(Published in Technology of Education, Vol. III, pp.2-5, 5/1995, Nitra/SK)
The study of artificial intelligence is closely coupled to that of natural intelligence. Thus, it is
not surprising that initial impulses originated from theoretical psychology and the realization that our brain is a
parallel-processing machine. It is obviously in the nature of intelligence, that the scientific exploration of AI and
its applications has spread over many disciplines in science and technology. In essence, the science of artificial intelligence has become multi- and inter-disciplinary. Active feedback between the study of natural and
man-made intelligence may someday, reveal its true nature. Under-standing the nature of intelligence may eventually lead to the design of intelligent and, perhaps, super-intelligent machines.
Recently, the term virtual reality has become a common buzz word. Indeed, computer-integrated
gadgets have been constructed that can create a pseudo environment where reality is no longer distinguishable
from its simulation. Although the implements of simulated reality have been used for many years, for example,
in flight and space simulators, the term, virtual reality, has become common-place only in the immediate past.
In this lecture, we will attempt to determine typical characteristics of intellectual behavior and will put
forth the proposition that intelligence, the natural and the machine-bound, can be virtual reality as well. An
analysis of intelligence and its manifestations, its universality and its elusiveness to definition will lead us to
the recognition that virtuality is reflected in many of our intellectual mechanisms, in our illusions and our
dreams, in our mental deseases and, perhaps, in our internal models of the world.
PROLOGUE
Intelligence is one of nature's most intriguing mysteries. It begins with its definition which is
equivalent to a physical entity, namely the brain, trying to explain itself, and it ends with our inability to completely define it. Defining intelligence is as elusive as our dreams, and depends often exclusively on our human
point of view. Intelligence manifests itself in many different forms, the most obvious being the human
mind that seems to emanate from the brain's architecture with trillions of neural cells combined to complex
networks. Criteria of intelligence are many with memory and storage of knowledge being apparently a necessary
prerequisite. Such memory can be hardwired (ROM-type) such as it is exemplified in the genes of our chromosomes which appear to exhibit deterministic intelligence, or softwired (RAM-type) as it shows up in our shortterm memory. Much has been learned by default about the brain, the mind, and its manifestation in what we
call intelligence. Specifically, the brain, diseased by drugs, by natural causes, or by accidental damage can provide us with important clues about the mechanisms that govern our intellectual behavior. In recent years, new
technologies have emerged that permit "insight" into our brain and its dynamics in relation to intellectual activities. Yet, a satisfactory explanation what intelligence is, appears to be eluding us.
Intelligence of some kind seems to be all-present: It is apparent in the single cell of the amoeba which
actively communicates with its environment if for no other reason than for survival. It is certainly obvious in
human brains whose superior (?) intelligence is often the cause of our arrogance: We are afraid to admit that
other types of intelligence could be possible, that of our dogs and that of machines. Whenever we stake out the
boundaries of intelligence, we frown upon a machine that can compute faster than we can do. Intelligence appears to be modified in societal structures that has its own dynamics and behavior: The masses in political rallies and street riots show (un)intelligent behavior that subordinates any individual intelligence.
This lecture is meant, on purpose, to be provocative and should challenge us to open our minds to the
secrets and intricacies of intelligence, the obvious and the speculative. For example, we will pose "crazy" questions such as, is intelligence, perhaps, only virtual reality, a fiction of our imagination?
What's it all about Intelligence: Defining the Essence
What is intelligence? This question has bothered humans probably ever since they started thinking.
By definition, intelligence is "the ability to learn or understand from experience, to acquire and retain knowledge, the mental ability". Yet, does this definition fully explain the essence? Also, without the physical architecture of the brain could there be intelligence?
Rather than dwelling on the pertinence of these definitions, we will list in the following, in random
order, what we perceive to be some basic ingredients of intelligence.
1.
Memory and knowledge, knowledge and know-how
Memory is the storage bin of knowledge. Memory in the human brain is still a mysterious affair.
Although much is known about the effect that memory has on our behavior and our being, we can only speculate about the physical structure and dynamics of it. Is memory, perhaps, some dynamic switching configuration
as we find it in the random access memory of the computers or is it a static imprint of abstracted signals that we
experience with ferro-magnetic domains in a recording tape? How does, perhaps, the chemistry of our brain control our intellectual behavior and our learning capacity? The analogy of our memory with the random access
memory of a computer is appropriate, since both the random access memory of a computer as well as the human
memory are gone as the machine is turned off or as the brain dies or becomes diseased. Also, the experience of
amnesia indicates that memory loss means often that access to our stored knowledge is blocked or damaged or,
in computers, that the table of contents is wiped out: The contents of memory are still there, yet hidden and
inaccessible. Thus, memory as well as access to it are essential to recall.
The phenomenon of memory, in a generous sense, can apparently be found everywhere: It is the carrier
of information such as in a cold front that still "remembers" the arctic temperature of its origin; it is in the flash
of electro-magnetic radiation telling us that a super-nova exploded in eons past. Memory seems to be "burnt"
into the genes of all living species which determine à priori the colors of our eyes, the features of our faces, and
the characters of our selves.
Without memory, there would be no knowledge, without knowledge, no know-how. A knowledgebased system then evolves from simple data-processing to information storage to knowledge which is defined as
"information that men or machines can think about". Thus, intelligence is a product of memory that is capable
of storing vast amounts of knowledge which is continuously modified through interaction between signals from
our environment, and our know-how as it is rooted in stored knowledge.
2.
Language and Signals; Communication skills
Intelligence would be absent without our ability to communicate with the surroundings. Indeed, we
would not know about intelligence without the capability to communicate. Communication is the process of
exchanging or imparting ideas, thoughts, opinions, or information by speech, writing, signs or other forms of
"Language": Language is the carrier of communication, and communication the vehicle for learning. Without language there would be no communication. Without communication there would be no intelligence. It has
been stipulated that language and intelligence are complementary to each other (See, for example, references in
W. Rudloff & C. Hering, 1989). Like the chicken and the egg question, we can only guess which one came
first. Language possibly evolved when humans made a conscientious effort to express feelings and ideas in
meaningful utterances, in sign language and sound. Language and intelligence have probably evolved in mutual
interaction with each other. In recent studies on chimpanzees, scientists have discovered that primates seem to
have already built into their brains the capabilities of understanding human language and can think in abstract
concepts (PBS-NOVA, 1994). Their problem is that their physical speech mechanism does not allow them to
communicate vocally as we do. Much could be said on how communication has been made possible between
men, between men and animals, between men and machines, and between machines through the evolution of
our natural languages and the development of artificial computer symbolics.
3.
Ability to learn, acquiring skills; Creativity and Intuition
One of the most outstanding features of intelligence is the ability to learn and acquire skills. Learning
is a life-long process whereby we interact with our environment, receive signals through our senses. We abstract
patterns from those signals, categorize the patterns while
comparing them with previous experiences, and store the
information in our memory as "knowledge" for future use.
The puzzling question of intelligence is that of
human creativity and intuition. Again the definitions of
these words tell us little: To create means, "to bring into
being, to cause to exist, to produce" or, "to evolve from
one's own thought or imagination". Perhaps, we should
consult quantum mechanics that allows for the tunneling
effect through activation energy barriers when atomic and
sub-atomic particles can have probabilistic presence on the
other side of the barrier where classical physics does not
allow it to be. Thus, in an esoteric sense, intuition is a
tunneling mechanism that permits humans to find the
solution to a problem without apparent effort (Figure 1).
Artificial neural networks have the ability to
learn by exposure. Indeed, it has been demonstrated that
Figure 1: Intuition and Tunneling
an artificial neural network can show similar behavior as
humans do during their learning process: They seem to be able to learn and forget in the process of acquiring
knowledge and recalling it (W. Rudloff & Bengtson, M., 1990,1991).
4.
Common Sense and Fuzzy Logic; the "Almost"-Gate
Common sense is another one of those buzz words that we all think we understand but that is difficult
to define. The American College Dictionary defines it as "sound practical sense; normal intelligence", yet what
is "normal" intelligence anyway? Again, we are confronted with vague expressions which, in essence, reflect our
grandiose ignorance. This is certainly a consequence of the physical dynamics of the neural system where the
simultaneous firing of thousands or millions of our neurons becomes a statistical affair.
While computers are designed around deterministic gates such as "AND", "OR", "NOR", and "XOR"
gates, Kent (R. Kent, 1981) stipulates an "ALMOST" gate that can accommodate the statistical behavior of our
neurons and may, perhaps, emulate the fuzziness of our thought process as manifested in our everyday language.
Logic and fuzzy logic are part of mathematical theory and have been integrated into computer science applications. Indeed, fuzzy logic, in combination with the concept of neural networks, has become a powerful tool in
artificial intelligence (B. Kosko, 1988).
It is intriguing to speculate how common sense could, perhaps, be emulated by a machine. For one,
memory plays an important part, but the memory is shared by social structures or “common”; shared with accepted or “realistic” concepts that have generally beeen agreed upon. Secondly, the notion of general acceptance
implies an overarching common consciousness where memory relates to a common pool of knowledge and a
shared or common model of the world.
5.
Symbolic representation and abstraction, comprehension
A major characteristic of intelligence is the ability to abstract meaning from simple signal sensing, and
from the complex signal transfer through the nervous system that seems to be a hybrid of electronic and biochemical mechanisms. One of the most striking ingredients of intelligence is the ability to abstract facts and
ideas into symbolic representations. Initially, sensory excitation of our neurons in the eyes, the ears, and other
senses is transmitted to networks in the brain where patterns are abstracted, recognized, and compared with pre-
viously memorized patterns. At higher levels such patterns are then further abstracted into symbolic form such
as words and sentences that have “meaning”.
In contrast, the fundamental symbolism of our computers is based on the binary language system
which reflects the on-off combinations of thousands and millions of electronic switches. It is this crude symbolism on which we construct more and more higher-level programming languages that allow us to talk to the
machines at almost the level of human communication. However, we still need compilers or interpreters to
translate the higher-level languages into the on-off switching states of the machine code.
6.
Logical Reasoning, Judgment; Problem Solving
Any living creature is faced with problems when dealing with the environment. We learn to cope with
problems when we grow up; a forever on-going activity that sharpens multi-dimensionally our intelligence.
Specifically, when we are young, our inborn curiosity leads us to explore the world through the receptiveness
of all our senses. Eventually, during our more structured learning process in school, we are taught to reason
logically, more one-dimensionally, i.e. we learn to arrive at conclusions within the quasi-mathematical framework of logic, the simple, true-false, yes-no, or binary logic that is also the basis of our computer language, and
the not so simple, the fuzzy logic developed only recently (L. Zadeh, 1984).
Problem solving can be just a mechanism of survival or a mean in itself when we play games of the
body and the mind. It is, thus, not surprising that computers have been "trained" to play chess and other games
at the master's level. Much has been written on games, their rules, empirical and otherwise and how they are
part of the AI concepts and their applications to game theory and its practice.
7.
Senses as Interfaces to the Environment: Perception, Pattern Recognition;
Surfaces and Interfaces
Our senses are the windows to the world, or phrased the other way around, they are the windows into
our brains. From a physics point of view, faces or interfaces are boundaries between two different phases. As
such, they are areas of defects. Indeed, the surface of a crystal lattice may be considered one large defect where the
interatomic or intermolecular bonding forces are unsaturated leading to reactions such as adsorption, chemisorption, catalysis, or crystallization. Surfaces are where the (re)action is: At the surfaces of semi-conductors, electrons and "positive holes" can move about through the defective lattice and give rise to catalysis and adsorption.
At the "surfaces" of hemoglobin, molecular oxygen adsorbs to be transported by the blood stream to areas in the
body where it is consumed in vital, energy-providing bio-chemical reactions. At the surface of our retina, electro-magnetic radiation interacts with specialized neurons to signal events from the environment which are subsequently transformed into nerve signals transmitted to the processing centers in the brain where patterns are
extracted, recognized and compared with patterns stored already in memory.
Unfortunately, the ranges of our senses are quite limited. The sensitivity of our vision, for example,
occupies only a narrow range within the electro-magnetic spectrum. Similarly, our olfactory sense, our sense for
smell, is very crude if compared to that of our dogs. Thus, quite literally, dogs have extra-sensory perception
that goes beyond the capacity of our senses. Potentially, we can build intelligent machines that can "see"
through clouds and have vision at night since the technologies of radar and infrared are well known. In the past,
as an exercise in futuristics, we have speculated on the possibility of super-intelligent machines that can incorporate such extra-sensory capabilities as part of their "intelligence" (W. Rudloff, 1987).
The importance of the senses to our intellectual behavior is reflected in deseases where all or part of
our senses are immobilized or even destroyed: Drunkards see white mice that are not there (virtual reality?) and
an amputated leg seems still to be attached as far as the brain senses it.
8. Intelligence is Connected to Physical (Living?) Entities: The Architecture of
the Brain; the Brain as a Computer
From all we know, any intelligence is
attached to some physical or material structure:
The brain is the physical (material) manifestation
of intelligence in humans. The question then
arises: Is intelligence possible in inanimate matter? Answering this question is the purpose of
artificial intelligence.
As a matter of evolution, the brain is
composed of three bio-computers (Restak, 1979),
one stacked on top of the other, and all three of
them are interconnected via the "busses" of nerve
strands (e.g. corpus calossum) which form the Figure 2: The Three Computers in the Brain
(Restak, 1979)
communication links between the three computers and the millions of neural networks as
processing centers of our knowledge (Figure 2).
The innermost computer, the R- or Reptilian Complex, is common to our brain and that of "lower" life forms
such as snakes and other reptiles. Here is where our instinctive reactions reside, the reactions against imminent
danger and the drive to satisfy our survival needs. On top of the R-Complex is the limbic system, the seat of
our emotions that so strongly influences our learning process (the stick and carrot effect).
Finally, the third computer is represented by the cortex, the outer shell of the brain. Here is where
highly intellectual activities take place, here is also where we see strong specialization and separation of control
of our actions as they are guided by the left versus the right hemisphere of the brain.
The fundamental building blocks of the brain are the neurons, and their cooperative effect in neural
networks are the physical manifestations of multiple parallel processing centers in the brain. The implication to
learning machines are on hand as we try to develop computers with parallel processing capabilities. I speculate
that the search for machine intelligence will get another boost when we succeed in building multiprocessors
with 1000 or, perhaps, a million cooperative processors placed on a single chip.
The human brain's architecture has apparently evolved over the millennia of life on earth and is manifested in the three major structures of the brain. One of the most remarkable observations is that the evolution of
the human brain is apparently replayed in time compression during the development of a human embryo: The
brain of an adult reptile corresponds to that of a seven weeks old human embryo. At eleven weeks in the embryonic stage the human brain is equivalent to that of an adult bird. Still later, it resembles that of a rabbit. Indeed,
at five weeks of age, the human embryo can hardly be distinguished from that of reptiles, birds or rabbits. Embryonic development is the time machine of evolution.
While the evolutionary process seems to have followed the road of trial and error -- our brain is still
encumbered with out-dated construction, the computer systems are designed in a more sophisticated manner. As
a consequence, they are always up-to-the-state-of-the-art, more efficient, and faster. We certainly would not
dream of connecting an advanced computer on top of an out-moded one, a highly sophisticated 486-based microcomputer with the old and slow dinosaur, the ancient IBM 650. This is where computers have a distinct
advantage over our brains: In our brains the antiquated machinery is still around and there is no way to replace
it with state-of-the-art equipment.
9.
Behavior and Emotions
Our behavior and the emotions are controlled by the brain and thus are part of our intelligent make-up.
As far as human and animal intelligence is concerned, behavior and emotions strongly interact with our intellectual processes. When we are depressed, our performance decreases; when we are in high spirits, we conquer the
world.
Again, we have learned about it through the default of desease and intrusive surgery: In past treatments, the corpus calossum that connects the parts of the brain, was severed to change violent beha-vior of patients. Today, we understand better that chemical imbalance is the reason for abnormal behavior. Through drugs
we can control our emotions and behavior; but more recently, we recognize that conscious interaction between
mind and body can release natural bio-chemicals to influence our emotions.
10.
How about Intelligence as Extra-Physical or -Material Phenomenon?
Extra-physical or extra-material intelligence, does it exist? We are severely limited in imagining beyond the three dimensions of space and, perhaps, the fourth dimension of time. Although we can derive the
mathematics of multi-dimensional space, we have trouble envisioning it. How could we visualize intelligence
beyond the fourth dimension? Is intelligence, perhaps, just another form of energy floating around and interacting with matter in accordance with Einstein's Theory of Relativity? The problem boils down to one of belief
versus scientific evidence. Pantheism of the ancient Greeks comes, perhaps, closest to an abstraction of a multidimensional intelligence where a godly intelligence is assumed to be distributed over everything in nature.
Epilogue: Is Intelligence Virtual Reality?
There are aspects of our intelligence that have virtual quality. Here again, we are facing the problem of
defining properly the terms. In the first place, virtual reality seems to be an oximoron. Virtual means “in effect
but not in fact”, i.e. not real! In the wonderworld of computers, virtual reality has evolved as a toy and as a tool
where reality is simulated with electronic chips, some headgear or other electro-mechanical devices. If we consider natural intelligence, there are clearly virtual aspects to it: Our dreams are certainly not real though they
appear so during our sleep. They are apparently the result of uncontrol-led (by our consciousness) firings of neurons where memory and low-level analysis of the processing brain cooperate to immitate reality mixed with
unreal yet believable activities.
Mental deseases and abnormalities such as schizzophrenia can create the illusion of several personalities in one brain. That’s virtual reality! Drugs like LSD seem to provoke an illustrious and colorfull pseudo
environment that is certainly not real. Often, natural phenomena play tricks on us and create a virtual reality.
One has to experience a mirage in the Rhub-al-Khali, the redish-colored sand desert on the Arabian peninsula,
to appreciate how one can get lost following the image of an ocean or an oasis as created by specific atmospheric
conditions. From all we know, UFOs are probably some kind of virtual reality where our imagination runs
away and creates an illusion of extra-terrestrials.
Finally, I like to argue that, perhaps, our intelligence per se is virtual reality where our surroundings
are reflected in our brain that processes environmental signals and interpretes the abstracted patterns in terms of
our internal models of the world that are, by their very nature, virtual.
References
Kent, E. W., "The Brains of Men and Machines", Byte/McGraw Hill, Peterborough, NH, 1981
Kosko, B., "Bi-directional Associative Memories", IEEE Transactions on Systems, Man, and Cybernetics,
vol. SMC-18, 49-60, 1988.
PBS, NOVA Series, February 1994
Restak, R. M., "The Brain: The Last Frontier", Warner Books, publ. by arrangement with Doubleday &
Co., Inc., New York, 1979
Rudloff, W. & M. Bengtson, "Phonetic Primitives and Their Potential to Automated Inter-Lingual Communication", paper presented at the 5th International Conference on Systems Research, Informatics, and Cybernetics, August 1990, Baden-Baden, Germany. Paper currently under revision for printing.
Rudloff, W. & E. Siebert, "A Proposition for a Binary-Coded Universal Knowledge Base: A Holistic Approach to Natural Language Processing", paper presented at the 6th International Conference on Systems
Research, Informatics, and Cybernetics, August 1992, Baden-Baden, Germany.
Rudloff, W. & J. Keigher, "Phonetic Primitives as Basis of a Universal Computer Code", in Adv. in Comp.
Sci., pp 54-59, 1989(A), IIAS Publication, Windsor, Ed. G.E. Lasker
Rudloff, W. & C. Hering, "Human Language, Contextual Understanding, and Computers as ManMachine Systems of Intelligent Knowledge Transfer", Transactions of the 12th International Congress on
Cybernetics, Namur, Belgium, August 1989(B)
Rudloff, W. "Beyond the Fifth Generation", Invited Plenary Lecture at the 1987 International Symposium
on Systems Research, Informatics, and Cybernetics, Baden-Baden, Germany, 1987.
ACKNOWLEDGMENT
This project was sponsored by a generous travel grant from GSU's Alumni Association under the
Directorship of Ms. Ginni Burghardt. We also appreciate the support that we received from Dr. Jane Wells,
Chairperson of the Science Division, and the GSU Administration.