Will computers ever take the power?

Part three.

Here in this part, we put aside the fantasy of a post-human futuristic society populated by a new type of "living being" in the form of highly advanced computers. Let's discuss computers as part of their current mission as machines created by man to automate calculations, information processing and control. And let's pose the question: how far will the process of computerization of large enterprises, or even cities or states, reach.

In other words, to what extent can the already existing role of computers assisting in design, research, control, and information services be expanded in the future? Can a moment come when a large production, a city or state will be entirely controlled by a computer in the COMPLETE absence of any personnel supervising them during a year or longer?

NO !

"NO" not because of the high cost of computers: they are now cheaper than dust.

"NO" not because of the still insufficient speed of computers: computers are already so powerful that their speed is more than sufficient for almost all administrative tasks.

The real reason for this "NO" is...

a) An increasingly expensive and difficult process of writing and debugging programs.

b) The high cost, unreliability, difficulty of setting up and maintaining the performance of servo-mechanisms and robots in automated enterprises that are controlled by computers. To maintain the performance of servo-mechanisms is a challenge even in the clean rooms of high-tech manufacturing – never mind the real-life environment in field deployment.

Let's discuss (a). Let's define the efficiency of a line of code of a program as the value

The number of all executed instructions including those in linearized loops

                  The number of written instructions in the code                               

The maximum is achieved in scientific problems based on mathematical algorithms and numerical methods. Say, a couple of thousand lines of such a program can make a computer to perform trillions of operations during a day, or a week, so that the efficiency of a line of code will be in a matter of trillions.

On the contrary, in programs that imitate human actions, business rules, and the actions of operators of complex processes the code is extremely inefficient consisting of thousands of steps such as

if ... then ... else ...

if ... then ... else ...

Efficiency in this case is just something close to 1 (rather than trillions). However, the worst thing is that such a program must reproduce all the conscious and subconscious actions of every competent specialist or manager whom it is intended to replace.

It is a pain to write such a program (hundreds of thousands of lines long) with the goal that nothing of the human knowledge and experience would be missed. And it is even more pain to find errors in it. Yet the utopian goal is to remove all people, from operators to the top manager, leaving the most crucial production only at the discretion of robots! Let's say a nuclear power plant!

Is it thinkable to send a passenger plane to a flight on autopilot from the start to finish, without any pilots on board (and without remote control of the aircraft by the operator)?! 

Yes, modern automated factories use many programs controlling robots and processes – but under the constant supervision of hundreds of specialists and adjusters of all robots. To replace all human adjusters and operators with the next generation of robots and servo mechanisms for them will again require even more skilled operators and adjusters! And so on, and so forth.

Thus, the fundamental obstacle to full automation of production and even wider facilities is the ever-increasing cost and unreliability of software, as well as the cost and unreliability of servo equipment. For these reasons, attempt of full automation will inevitably hit the ceiling in a process of trying to make it more and more comprehensive, never reaching full automation. At the same time, the catastrophic consequences of inevitable automation errors will also increase.

Part one.

As it has just been analyzed, there are fundamental limitations in attempts to widen the use of computers even as assisting machines in large-scale production. Then is it reasonable to talk about a post-human society of the future, in which computers have replaced people, if there is a ceiling even in the automation of production and services?!

However, the most important thing here is that computers are man-made inanimate machines that execute only the algorithms entered by man. Computers are the most typical example of a completely deterministic physical device implementing computable processes – and therefore they cannot model consciousness (see the ideas by Penrose below), nor to have its own "goal setting" or "free will". Computers can do that and only that which was entered as a program by a human operator.

Of course, an evil operator can create a program that controls the servo-mechanisms which will start killing people and "taking power." However, this would mean that it was not computers taking power, but the person with his accomplices who armed himself with computers (and various murder weapons) to usurp power.

Talking about the taking power by "beings" incapable of goal-setting and free will is SENSELESS.

If we however imagine a new kind of computerized person as a living human brain inserted and functioning in a robotic body with a computer, then the soul of such a creature is the same as that of this brain (once it was part of a full-fledged person). Such a robotic man will indeed retain the ability of goal-setting and free will, plus he will have great physical strength and computing abilities. But how is such a symbiosis better than a normal person using computers plus mechanical amplifiers? Such a symbiotic robotic brain is, in fact, the most severely disabled person, equipped with prostheses. He still needs normal physical conditions and a blood supply of renewed blood to survive. He, just like a man, is not suitable either for flights to planets, or, even less so, for flights to stars (being a "piece" of matter of non-zero mass). What are the super qualities of such a creature? It is pointless to fantasize about the taking of power by this kind of severely disabled individuals.

Part two

And now let's consider the question of whether a computer, as an absolutely deterministic machine without its own goal-setting and free will, can at least create the illusion of a soul, i.e. to model the properties of the soul and consciousness in full.

In full – no! A number of scientists such as Penrose, Hofstadter and others prove that the functioning of consciousness (and subconsciousness) cannot be modeled by any algorithm.

However, even putting aside such a mysterious process as human consciousness, it has long been proven in the theory of algorithms that a number of seemingly "ordinary" and useful problems of mathematics proper and the theory of algorithms cannot be solved by any algorithm. For example, there is no algorithm that can determine whether a given algorithm in question will ever stop. I.e., even in such a narrow area as solving problems in mathematics and the theory of algorithms, the abilities of algorithms have a ceiling.

Algorithms can create an illusion of non-deterministic behavior: for example, an algorithm that generates pseudo-random numbers. A sequence of pseudo-random numbers has exactly the same mathematical properties as a sequence of real random numbers (though the sequence of pseudo-random numbers is deterministic and reproducible).

One can insert into the computer a physical device that generates physical random variables. In this case, the sequences of random numbers will be irreproducible, but the process itself, being physical, is deterministic by definition, therefore, there is no free will.

Prologue

Analyzing the fantastic assumption "whether computers will ever take power" over people, one must clearly realize how a "computerized person" or "computerized higher living being" differs from a human.

According to our definition, the human, a "supreme living being" is one having an immaterial soul, i.e. a non-physical substance, outside the physical world and outside the physical laws, capable of goal-setting and free will (and also feelings); having the soul whose functioning (according to Penrose) is incomputable process, i.e. a process which cannot be modelled by any finite algorithm. Because of this alone…

No computer program and no computer may be viewed as a human, so that the expression "computerized person" or "computerized higher living being"  is an oxymoron.

 The Turing test is unable to detect whether the creature in question has the ability of goal-setting and free will! The Turing test only detects the semantic and literary quality of the speech generator – in the perception of these qualities by the person conducting the test.

Programs like "Talkative Eliza" written in plain Basic already worked well 30 years ago. Now programs like Copilot, Gemini, etc., using AI algorithms and huge databases of information on any topic, are already capable of not only conducting a "human" dialogue with the user, but also detecting its meaning and answering complex scientific questions at the level of an average living person.

Then what is the test for the presence of goal-setting and free will? (At that, the feelings of a given soul, by definition, cannot be detected by someone or something outside).

And here comes the bad news. Even a living person on his own cannot logically prove to himself whether he has free will.

He can accept the axiom that he has free will – and no experiment will disprove it.

And he can accept the axiom that "free will" is merely his own illusion, an interplay of subtle physical effects of his purely physical deterministic body (while there is no soul at all). And again, no experiment will disprove it.

But then no experiment would disprove solipsism either (if one believes it)! The experiment that the hero of the Solaris novel staged on himself does not actually prove that he is not dreaming.

However, all these uncertainties and doubts only apply to humans, to the highest living beings known to us on earth, and to hypothetical "creatures" in question on other planets. With regard to computers, there is no uncertainty, because they are special machines created by man to perform absolutely deterministic actions.

Therefore, speculations about a possible takeover of power by computers in remote future are unfounded.

Alexander Gofen

2022-2024