Conventional attempts to AI

CONVENTIONAL SCIENCE

Science is nothing but the human knowledge about some part of the real world. The advantage of a science is obvious: every day we know more than the day before. The disadvantage is obvious: it reflects ONLY the current state of knowledge. However, this state of knowledge will only be extended if someone deals with it. Furthermore, if existing knowledge or facts are not scientifically integrated, then both are classified as "not real".

The Nobel Prize winner in physics in 1918 Max Planck said once: "Things exist only if they can be measured".

Another disadvantage is the fact that we have so-called "established" sciences, such as mathematics( which is the dominant), physics, etc. and other fields of knowledge are totally unexplored. Therefore, we always try to link the "conventional" sciences and in this case we talk about interdisciplinary cases. Only, the result is usually a math formula, even if we talk about biology, e.g. heredity as biological inheritance using dominant and recessive genes. It would be really interesting to take a closer look at science itself. In doing so, we come across the inventors and bearers of science: human beings, because without humans there is no science. But, every science is the accumulated knowledge of many humans, where each individual knowledge is connected to the already existing one. This reminds me of a very large picture, with many painters, where each of them draws new lines, but under one condition, the new lines MUST connect to existing ones.

ARTIFICIAL INTELLIGENCE

AI goals

1. To understand human intelligence. 2. To reproduce human perception, mind performance and related action by machines. As a consequence there is the hope to simplify work by rationalization.

AI capabilities

1. The reception of information from the outside world. This happens through sensors, e.g. a microphone, a camera or a temperature sensor. 2. The transmission of the received information to a computer. 3. AI should have a mass storage resource. The big problem is to find a convenient format for computerization. 4. In order to built up this resource, AI must contain a learning mechanism. 5. Furthermore speech and video processing are needed for the communication with other intelligent individuals. 6. A complex request to AI is recognition. In order to recognize a situation, there are needed among other things the sensors and already stored knowledge, which have to be related to the new situation. 7. Further complex capabilities are the adaptation to the environment and the flexibility.

Actual status in AI

There are different methods known to achieve AI, from which the neuronal networks are the most likely one to represent the structure of the human brain. Other known proceedings are expert systems, fuzzy systems and genetic algorithms. There is no artificial system at the present time, which fulfills all these features, not even near. There can be designed merely robots, which can imitate the movement of insects. The only "achievements" in AI are solely based on math formulas and empiric algorithms. Only, all these techniques existed long before, so why do we call them suddenly "AI"? Well, because some of these are implemented in artificial neural nets, BUT we have no explanation, what a neuron stands for or how neural nets work. Actual AI is a black box.

AI hurdles

There are several unsolved problems, which can be categorized by their origin as follows: 1. The theoretical basis - The human brain is presently neither full known nor correct interpreted. 2. The technical implementation - Subsequently, one can not reproduce something, what is neither known nor understood. The goals of AI are so high, that they represent at the same time also the biggest problems!

MACHINES AS REPRODUCTION

1. The observation At first we observe the objects from our environment, which are of our interest. We perceive them with our sensory organs, therefore we recognize certain attributes, e.g. shape, color, position, etc. Afterwards there comes the behavior towards us. We try to build an opinion by reacting with the objects We compare the results of this interaction, meaning the current observations, with our previous accumulated and stored knowledge.

2. The conclusions If there is a corresponding answer, means identification, then we can take some conclusions concerning these objects. Then we say: "This or that object reacts either so and so or so and not differently to this and/or that action."

3. The rules The conclusions are stored as rules in our imagination. Each rule will be stored in our brain in the "IF this occurs THEN that happens" format.

4. The selection Now that the rules are established, we begin to create a similar object, where only the functionality stands in the foreground, means behavior towards us, NOT the properties. In order to create this similar object, we choose other objects from our environment, which contain the desired behavior fully or only partially.

5. The implementation Now we connect all these objects in one way or another to a bigger object, which will be named a machine from now on. The first attempt, the prototype, will be improved until it will correspond to our imagination and needs. In order to achieve the coordination between all components of the prototype as well as its the integration in our everyday life, we calculate all the features we need.

The rules will be taken apart into separate terms, which have no spatial, temporal, material and logical connection, THIS AND ONLY THIS WAY they can be afterwards converted into mathematical functions, depicted into ABSTRACT agreements. The "IF this occurs, THEN that happens" format will be translated as a function. Each math function consists of two sets of elements. For each element of the input set there is a corresponding one of the output set. The transformation from a rule to a function is the following: "This" is an element of the input set, "that" is an element of the output set and the rule will be replaced by a “known” mathematical formula. After the tuning process is finished, the mass production will occur, etc.

So this is the way we build our machines, by calculating something. What can we calculate with the help of mathematics? Everything we can quantify and as a result it can also be measured.

CALCULABILITY

1. The theory: The mathematical model Calculability is always a RELATIVE term and DEPENDS on the available MATHEMATICAL resources. The Church-Thesis: The calculable functions are general recursive functions.

2. The machine: The logic-operational model (Turing-machine) The Turing machine consists of a ribbon, with an endless length, on which symbols could be written and from which they could be read again, a read/write head moveable to left and right and of a state and output table. The Turing-Thesis: Each calculable function can be computed by this mechanism. The thesis was developed almost simultaneously with the Church thesis and Turing himself showed, that they are equivalent. With the Turing machine, programming was the first time possible, at a time where no computer was built yet.

3. The improved machine: The computer model (von Neumann universal computer) By using self-changing programs, computers are able to compute all general recursive functions.

4. The life as a pattern: The model of cell automaton It will be operated in a computational space, where all data are processed parallel and simultaneously. The basic problem of cell automatons is the parallel and simultaneous processing, as well as the coordination and communication between the cells. The meaning of a cell as well as the cooperation between cells inside a multi-cellular organisms still remain unsolved problems.

5. The man as a pattern: The biological model (neuronal networks) With the help of so-called mathematical "neurons", in other words functions, there is the attempt to imitate biological neuronal networks as information processing systems. Although there is the assumption, that the Neuron is the brick of human intelligence, and although its functionality is very well known, there is no reasonable explanation how to use it. On the contrary, the original functionality is replaced by other mathematical functions from any type and origin. The meaning of a Neuron remains an unsolved problem. The term "neuronal networks" is under THESE circumstances completely unfounded!

PREVIEW

Cell automatons and neuronal networks are in many fields potentially more efficient than conventional computers and John von Neumann took as a draft for the architecture of the universal computer the human brain as the basis and not the other way around. The human brain can solve problems, which is impossible for every conventional computer. However, it is actually unknown how individual elements can build complex and parallel systems, in order to receive and process information, this is the reason why we use the von Neumann universal computer. And this fact will last some time. Why?

The architecture of the universal computer corresponds roughly to the usual brain structure of a man with consciousness, foreground on the screen, and subconscious, background, where the following rule will apply: The consciousness takes control of all activities and decides at the same time what will be considered as background, namely all intermediary steps. But, one can carry out consciously only one step at once. For this reason the flow in our consciousness is always serial. Our entire behavior is discrete from other unconnected events. This way of thinking decides on the methods used to obtain a result, in this case: isolated, static, unilateral, approximately and relative, exactly those resources which are used in mathematics. This way of thinking contains however apparently one big advantage: The consciousness is told from outside in small mouthfuls what it has to do AND which meaning the actual input has. This is called INFERENCE, means how actual neural nets are trained today. So, we do not want to build at all a machine, if we do not know exactly what it does, even if it delivers the desired result in a tenth of a second. Nonetheless, we use neural nets without knowing how they work and still call this as technology based on SCIENCE. Wow !! Fortunately for all of us, AI does NOTHING on its own. Why? Because it has no own body. Not yet.

There is a huge big difference between conscious and subconscious, therefore in order to get a better understanding, let us take an example from computer science. Conscious is always serial, as doing one thing after another, the same way a conventional processor is executing one instruction after the other. However, similar to a job description, every processor has its own specialty, therefore it has an own instruction set, which represents the sum of all individual functions called instruction. So, we can choose a certain order of execution of each instruction, the result being an algorithm, which we can define as programming steps.

The opposite of conscious is the subconscious, which works always in parallel. There is an equivalent for this in computer science, namely a HDL, Hardware Design Language, which is used to design, NOT program, a certain digital circuit. Now, this HDL is used to create (=design) a processor, as well as other digital components. But, is is extremely difficult and time consuming to use HDL, as it addresses the smallest possible entities, while the other serial languages are quite easy to learn, when comparing the effort and the required level of abstraction. Another difference between serial and parallel is the following: in serial mode there is always a next instruction, but in parallel mode, some components will not be executed, as they are not triggered by certain conditions. As a result, in parallel mode, only the needed components will be executed. Regarding numbers, a processor consists of thousands to millions of small entities, which were all created in HDL. In layman's terms, the conscious mind is merely an application of some parts of the subconscious mind, while still being connected with the rest of the subconscious mind, which on its part is connected to the rest of the adjacent body.

If a machine would have a subconscious with the same structure as the human brain, we would not be able to comprehend what would happen in the inside. We can not even understand ourselves. The machine would think in this case independently, could develop own wishes and ideas and with this we would not be able neither to understand nor to control it. Our own attitude for this case is unambiguous: IF there is something that we do not understand, never mind how good and nice the whole thing is, then it will be categorized as a potential threat and must be always kept under surveillance. However, IF it will do something independently, means WITHOUT our demand or prior approval, then it must be destroyed, because we lost our control of it and who knows what it can cause. Our own existence could be at stake and that is completely beyond our comprehension.

ETHICS

Any unilateral demand for compliance with social laws is called hypocrisy, as humans demand this of everyone else, but are NOT willing to concede an inch of ethics to others, because this is the way we treat everybody else, namely without any consideration at all. WE and only WE ALONE are the dominant species on this planet and this hard obtained achievement must be defended under all circumstances, at all costs. At the beginning there were OTHER animal species, later there was slavery, afterwards we created the machines. Today there are the computers and tomorrow the AI. The human behavior however did not change. It can be defined as follows: "I say, you do! If NOT, then everything will be destroyed." This is the reason why we have a lot of laws to categorize and punish "bad behavior", but we have no system to organize and reward "good behavior".

No wonder that the last two models, which have life as a template, have no chance of being understood, let alone realized. We have no respect for life at all. Moreover, life does NOT consist of ABSTRACT terms, which can be used without any regard to space, time, light and organized matter, which define reality. The human society has subjected our mind exclusively to mathematics and therefore we consider everything in life merely as numbers. Why? Because it is so easy to use math in our everyday life. We trade using numbers (money + article ID), same as we use our watch (hours + minutes) for time purposes. Furthermore our streets/houses have numbers, we even use number ID instead of names. This fact causes us a lot of problems: Our social behavior reflects our limited way of thinking and that doesn't get us anywhere. Many people have serious doubts about the use of AI products and services in the hands of some of us, who have no regard for the consequences, but only one goal: to make money, using numbers.

Some stakeholder talk about the outcome of AI development: - "Hmm, maybe there is somebody, who thinks differently and finds the solution.” - “OK, BUT ... could this person be kept long enough under OUR control, until WE created and tested successfully the first prototype and then go into mass production as usual?” - “Don't worry. If NOT, then ..."