This article is not about the usefulness of artificial intelligence (AI), but an attempt to address a more profound question that has to do with the very nature and meaning of human life in the view of AI.
Saying that AI will never replace true humanity is not to say that AI system isn’t useful. In fact, AI is not only useful, but will become far more useful in the future than most people can imagine today. These are fundamentally different things.
“The Kurzweil View” and “The Gilder View” of AI
People’s view of artificial intelligence (AI), whether they’re conscious of the view or not, can be categorized into two different types: “The Kurzweil View” and “The Gilder View”, represented respectively by Ray Kurzweil and George Gilder, the two greatest minds of our time, except that one happens to be wrong and the other right.
Kurzweil works for Google and is one of the most famous futurists in the present time, well known for his making popular John von Neumann’s idea of “singularity”, while Gilder is an independent thinker, known for its accurate predictions of several major paradigm shifts in technologies put forth in his books such as “Life after TV” and “Life after Google”.
The Kurzweil View is wrong, not just because it is morally wrong, but is, and will be proven to be, objectively (scientifically) wrong.
The Gilder View is correct, because it rightfully perceives the true nature of human life in its relationship with the universe and its Creator, is predicated on a proper understanding of “spirit, soul and body” constituents and the purpose of human life (see below), and further supported by Gödel’s incompleteness theorems (see below). Besides, it is morally right.
Spirit, Soul, and Body
Man is “spirit, soul, and body” (1 Thessalonians 5:23).
The body is completely “self” or “eigen“, while the spirit is fundamentally extrinsic to the material (external of human existence altogether, not merely external to each other).
The soul, however, is in the middle, being not only a result of the interfacing between the spirit and the body, but also the interface itself.
The Kurzweil View not only denies the existence of spirit, but further treats “soul” as a pure function of the body, which includes brain, and therefore can be completely programmed. Based upon this assumption, it is only natural to conclude that human life is just a primitive form of intelligence and will soon become inferior to artificial intelligence.
In other words, the Kurzweil View does not elevate AI to the same kind as human intelligence, but rather degrades human intelligence to the same kind as a machine function.
If there is a Creator of the universe and mankind, such degradation would be an utmost disrespect and ungratefulness to the Creator. But because people who accept Kurzweil View, at least those who consciously do so, tend not to believe in the Creator, this is merely academic to them.
The information theory, however, suggests that human creativity and inspirations come not from within but from without, because mathematically it cannot come from within if human brain is considered a self-consistent system (see Gödel’s incompleteness theorems below).
Gödel’s incompleteness theorems
Gödel’s incompleteness theorems were revealed by Kurt Gödel in 1931. Gödel’s First Incompleteness Theorem may be summarized as follows:
“Any consistent formal system within which a certain amount of elementary arithmetic can be carried out is incomplete; i.e., there are statements of the language of the formal system which can neither be proved nor disproved in the formal system.”
Short and plain version: a consistent formal system cannot be also complete.
Here, “completeness” means the system is always self-explainable, that is, everything comes out of the logic (language) of the system can be explained (derivable) within the system itself.
For simplicity, every algorithmic system such as computer systems is a “formal system”. Artificial intelligence is such an algorithmic computer system, and therefore AI cannot be both consistent and complete according to Gödel’s incompleteness theorems.
These proven theorems have profound implications to the most profound questions of the universe. But people do not agree on such implications, and the cause of differences are not academic but all because different people have different axiomatic assumptions of the relationship between Man and God.
Some think that Gödel’s theorems have proven the existence of God. But in fact, Gödel’s theorems did not prove the existence of God, nor did it prove otherwise.
Let me repeat: Gödel’s theorems did not prove the existence of God, nor did it prove otherwise. It ought to be the case, because the notion that the Creator would allow Himself to be strictly mathematically proven or disproven offends the very character of God that has been revealed in the Bible.
But implications of these theorems become extremely powerful when they are placed in the context of the following statement:
“If there is no God, then Man’s mind (brain) is no more than a machine. But conversely, if the Man’s mind (brain) is more than a machine, one will have to acknowledge a fingerprint of God on Man in order not to violate Gödel’s incompleteness theorems.”
The first half of the conclusion part of the above statement can be called “mechanism”. But the opposite second half points to theism. This is not what Gödel said or even meant, but it is a necessary implication that has enormous significance.
It is from the above big “if’s” where people diverge.
Gödel himself pounded upon this question, as he stated:
“either … the human mind (even within the realm of pure mathematics) infinitely surpasses the power of any finite machine, or else there exist absolutely unsolvable diophantine problems.” Gödel, 1951.
British philosopher J. R. Lucas, refusing to accept human brain being viewed as a mere machine, made the following arguments against mechanism:
“For any formalized system, or a finite machine, there exists the Gödel sentence which is unprovable in that system, but which the human mind can see to be true.”
Lucas’ argument for the superiority of human mind (brain) is rather inspiring, but problem is that, although we generally do know human mind can perceive truth the way that a machine does not, Lucas’ argument, if viewed under the same mathematical framework of Gödel, is precisely “unprovable“.
Therefore, it inevitably goes beyond academics and comes upon faith.
The key is, how do you see the relationship between Man and God?
If you do not believe God exists, then Man is alone, and the mind is objectively no more than a Gödelian formal system (i.e., just a type of machine, and nothing fundamentally unique about him, or it). Then the dilemma arises from this view is that many people, like Lucas, intuitively know that Man does see something that machine does not. If one feels this way, but at the same time thinks like there is no God, his only honest choice is to write off his intuition as something that is too primitive and too self-centered to be regarded as a worthy vessel leading to truth. As he does this, his atheism belief or argument remains dignified, and the debate can not only go on with the vigor, but in fact victoriously so as perceived by himself.
But can we please consider the alternative too?
What if God exists, and God is not part of a “formal system” and human logic, and therefore is not subject to Gödel theorems? And, what if, as designed by his Creator, the Man’s mind is something uniquely placed somewhere in between a “formal system” and God, such that it routinely performs like a machine, but is interfaced with “external inspirations” that lead to perception of real truth and even real creativity?
By “real truth”, I mean something far more than information mapping but rather actual physical embodiment and connection with the truth in the universe (see below). And by “real creativity”, I mean something that actually creates true new information that is not part of the baseline machine realm of the brain (also see below).
More specifically, the biblical view would tell us creativity and inspiration originate not from the body (brain) but from the spirit, and is processed through the soul, and materialized with the body (brain). The fact that the human brain has amazing computational power is a necessity for creativity, but not the intrinsic reason for creativity.
According to information theory, any creativity, if it represents creation of true new information, is always a surprise (the source of which has to be external to a programmed system), and cannot be a result of a programed system.
AI cannot do this. Just because new AI algorithms already involve processes and results that go beyond what the programmers can rationally explain does not mean AI is creating true new information in its strict definition according to information science. Everyone knows everything is predetermined, albeit unexplained. To overcome this, programmers started to add randomness into AI. But introducing randomness into a program is not an equivalence of creation of new information, but simply an artificial way to introduce uncertainty or even fault to disguise the program’s inability to create new information.
It is because of this reason that AI may outperform human being in many functions, but will never actually become creative, and will not have a soul, much less a spirit.
People seem to have forgotten Gödel’s incompleteness theorems today. Gödel wasn’t a mere theory or hypothesis, but a fundamental mathematical theorem.
The most fundamental belief (and its biggest fallacy) of humanism is its axiomatic (even unconscious) assumption that there is in nature a closed and independent system that is fundamentally “self explainable.” Under humanism, first it was the universe itself that was placed in this category of “self explainable systems,” and later the human mind, and now an AI system.
But it cannot be true if Gödel is right. And of course he’s right. It is mathematically proven.
This fallacy is the root of all spiritual and mental idolatry, because it is making a god out of nature. Only God, the Creator, the “I AM I AM” (Exodus 3:14), is self explainable, because God is not part of the nature subject to Gödel’s incompleteness theorems.
Ironically, when this notion of “self explainable” human mind is challenged by Gödel’s incompleteness theorems, the same proud humanism quickly retreats to the opposite side, into believing that human mind is nothing more than a machine, and therefore will be replaced by AI. Humanism devotees of course did not give up on the pride. They found their humanistic pride in AI.
“Mapping” and “Meaning”
People who claim that AI has, or at least will eventually attain, humanlike understanding, are confusing “Mapping” with “Meaning”.
“Mapping” is to map information to a set of definitions, while “Meaning” is associating (or embodying) definitions and combination of definitions with the reality. An AI-powered computer can do a better and better job in mapping, but will never be able to actually understand meaning.
For example, seeing an apple falling from a tree, and further being able to mechanically connect the picture with concepts such as “apple”, “tree”, “fall”, and even deeper concepts such as “speed” and “acceleration”, are all acts of “mapping”, not “meaning”.
But appreciating the scene as something related to the observer’s human experience by invoking the observer’s whole being’s physical connection and perception of the real world, is “meaning“. And further understanding that the experience is connected to a universal law would be a unique human intelligence. And even further having a stroke of genius that gives the insight of gravity as an abstract concept which perceives and appreciates the beauty of the governance of the physical universe by a unified law and order, is “creativity“.
A machine does not attain any of these levels, not because it presently does not have enough computational power, but because of how it is connected to the real physical universe by design. See below for further discussion.
Once such meaning is expressed, concepts would have surfaced at the language level and become explicitly expressible definitions, it then becomes “mappable”, but such subsequent mapping does not equal to understanding the meaning itself, because meaning is not only connecting information to definitions, but actually associating (embodying) definitions with the reality.
Sometimes people support this correct view by arguing that the reason why the computer will not be able to do what human brain does is because the human brain has vastly superior computing power. A fact often cited is that a typical human brain has more than 100 billion neurons, much superior to today’s AI algorithm’s artificial networks which typically have 10-1000 ‘neurons’ per layer, and a maximum of 100,000 ‘neurons’ in a model that has many layers. (Note: the number of ‘neurons’ in an AI model is not merely limited by computation capacity which increases exponentially with the number of ‘neurons’, but also limited by the very nature of the AI algorithms which have a threshold of maximum neurons beyond which the training result does not improve but deteriorates.)
That is quite dramatic and convincing. But the truth is that the superiority of the human brain to AI is not even that. It is much more dramatic, not just quantitatively but also qualitatively.
First, the number of neurons is only relevant to “mapping”, the first and a superficial part of the “computation”, not the actual “meaning”. When it comes to meaning, it is not about merely mapping and deciding on probabilities. It is about actually embodying definitions with the physical reality.
Here, “reality” refers to the actual physical universe which is the ontological reality itself.
To understand this, one must remember the human brain is part of the universe, and is of the same nature of the universe, because it is made of the same physical material, such as protons, neutrons, electrons and photons, etc. which obey the same physics law, and is further designed to interact with the original purpose (i.e. not a simulation with redirected purpose). Therefore the brain speaks the same “language” as the universe, and it intrinsically “understands” the universe. This intrinsic understanding of the universe does not need to be scientifically expressed. That is, its existence does not depend on scientific explanations. It just does, and is.
But, if you really want to force a scientific explanation, you can start with quantum physics. The elemental particles in your brain react with each other and react with the external signals the way they do because they just do, by their very existence, not because they have a theory about it.
And if you compare the physical reality with the artificial computational power, you should quickly grasp part of the reality: there is absolutely no comparison.
For example, take the simplest chemical reaction in which a hydrogen atom meets another hydrogen atom to form a hydrogen molecule. In quantum physics, it is a matter of solving Schrödinger’s equation that involves four moving elemental particles, including a proton and an electron in each hydrogen atom. In the physical reality, when a hydrogen atom meets another one, they simply form a hydrogen molecule. They do not need to do calculations in order to guide the process. The nature has its own law that rules the process. It’s guaranteed. But if you use a modern computer to solve the Schrödinger’s equation, you will find even the simplest quantum process is beyond a strict analytical solution. Rather, approximations have to be taken.
How much approximation? Once you move beyond the simplest two atom reaction, the complexity increases exponentially, and more approximation is required.
Do you have an idea how many atoms your brain has? You don’t need to get into a strict calculation. Just a rough estimate will give you an idea. It all relates to Avogadro’s number, 6 ×10²³. This number tells us how far our real macro world is scaled away from the micro world. Avogadro’s number means that, for every gram of hydrogen, there would be that many (6 ×10²³) hydrogen atoms. The measurement for other matters differ, but they’re all on the same scale. And your brain has more than 1000 grams of matter, an additional three orders of magnitude.
Now that’s the 26th power of ten, not 26 times. That is billions of billions of billion times.
The point is that, when your brain is experiencing something about the reality, its complete experience is absolutely not possible to be wholly simulated using a computer, not even just purely from a materialistic point of view (which is not even true, see below).
You might say that computer chips are also made of the same natural material, such as protons, neutrons and electrons etc., so how they are different from the human brain in this respect?
It is different because the brain is designed to interact with the outside with an original purpose that is connected to the human soul and spirit, while the computer is designed for a simulation with a redirected artificial purpose.
The simulation purpose always starts with a human understanding, such as a scientific equation (Schrödinger’s equation, for example) or a statistical model in order to perform the computational logic necessary for the simulation of that human understanding. But the materials in computer chips are structured as devices (transistors, gates, conductors etc.) which are fundamentally different from the natural material in brain of a human being. These devices are not part of a pre-existing consciousness, but are simply computational devices that each is designed to extract a bit of information at any given time from a bulk of material existence which contains vastly larger quantity of physical information, which is unused in computation.
This character is not a defect of a computer, but rather the biggest achievement in computer science. It made it possible to create a computational process that is not only understandable but also controllable, repeatable and improvable by human being.
From an information science point of view, it transforms a natural abundance (silicon atoms) to an artificial scarcity (computational information).
The computer is not alive as a human being, but only “computes” according to an artificial logic. From a computer science point of view, it is absolutely a great trade-off because it has made certain things possible that were originally impossible. But one cannot claim it is like a human brain performing its original purpose with every natural atom participating at the atomic level, much less to claim the silicon atoms are somehow connected to, or able to create, a consciousness.
Just because a live human brain computes does not mean that anything computes is comparable to a live human brain. This is the case even if computers can vastly outperform human brain in computation.
The Unsimulatable Human Experience
Computer simulation aims at a certain aspect of the observable behavior, and can achieve highly usable results by using approximation. But just because a narrowly defined parameter of an observable behavior can be approximated well does not mean your computer is matching the actual physical experience of your brain.
Moreover, human experience is beyond the brain itself. The embodied cognition theory has taught us that our whole body is part of the experience, including learning.
It is that kind of physical experience of the universe that defines the meaning of human understanding and experience, because your brain does not just do mapping of definitions, but has an understanding of the actual meaning by associating (or embodying) definitions with the reality.
For example, when you speak the word “joy”, a computer that has voice recognition may use a better and better algorithm to accurately map the sound of word “joy” with the word “joy” in vocabulary; and if algorithm has really good AI, it may even map the word with a variety of phrases and expressions to create a sense of context, but only your brain understands the meaning of “joy” because all your being, every molecule in your body, has experienced that concept, and associates the experience with the reality in the universe.
AI is at most a simulation of certain superficial aspects of the experience. The approximation is not 10 times or not even a billion times simplified from the actual human experience, not even by a factor of Avogadro’s number, 6 ×10²³, because the complexity of the universe (including human body) is not linear to the number of particles, but exponential. There is a network effect in reality.
If we approach this with humility, we can figure out a lot of useful ways to develop AI that helps us to do things that we as human being are not good at, such as memory, mapping, calculations, etc. But to think that AI can actually replace human experience and meaning is beyond being foolish. It is ungodly arrogance, and babelesque pride.
And the above is just the materialistic side of the story of human experience. As said above, human experience is beyond the body’s physical experience. It connects to the universe not merely by physical particles (material), but through the soul and the spirit, which the machines simply do not have and will not have in the future.
Perhaps one will ask, what about man-made humans in the future?
But ask yourself this question first, how are humans going to make humans? They have done so using God-given reproduction methods in the entire human history so far. Are humans going to synthesize humans in the future? That would be evil, but if they further claim that they have actually made humans just like God did, it would not only be evil, but utmost disgrace and ungratefulness, because they would have just usurped the power of God-given natural law to try to mock God. One can trust that God, if God at all, cannot fail to provide a solution or judgment to this problem.