Scampish notes.
I will publish some of my long-standing notes on the topic of AI in the form of "as is".
12*. From the concept of Alan Turing, “if the interlocutor cannot determine whether he is talking to a person or a machine, this proves that the machine behaves as smart as a person”, it follows that if a PC with a free conversation with a person yourself as a person, then the PC has intelligence.
Not every person can talk to any person on any topic. For example, not every programmer-specialist in the field of any language will be able to support a conversation with a scientist biologist on the topic of “reproduction of ciliates of shoes”. In everyday life, people often cannot find common topics for conversation ... etc.
')
Any person in the workplace is currently little different from a “computer”, i.e. from a computer with a specialized program to perform certain functions. For example, when a non-standard task appears, employees who do not have information about the methods of its solution, as a rule, pass it on to other employees, at a level higher or lower. The program also gives an error message. That is, for a person who orders a ticket via the Internet and buys it at the airport’s ticket office, the difference between a person and a computer is insignificant, and if you add certain functions to the program, or select a specific cashier at the airport, the person can no longer distinguish the person from the program ...
Also, “conversation” can be interpreted as communication, the transmission of information which is owned by the interlocutors.
But this includes the following aspects.
Suppose there are several categories of information. And each subsequent category of information is nested in the previous one, and in order to process it, you must have access to top-level information.
Example:
1. Section.
Category.
Article.
Those. to read the article, you need to get access to the category, to get access to the category, you need to get access to the section.
Take also the "outlines of consciousness" Leary - Wilson.
1. Biovietal (unicellular) consciousness. Characterized by the perception, processing and transmission of nerve signals responsible for cellular health and reproduction. Metabolic safety.
2. Emotional-motor consciousness (consciousness of mammals). It is characterized by the perception, processing and transmission of neuromuscular signals responsible for the mobility of the body and control over the territory. (Fight or run?)
3. Symbolic-artifact consciousness. Perception, integration and transmission of information using gestures and abstract symbols (first alphabet). Creating language and tools. Symbolic thinking.
4. Industrial consciousness. Perception, integration and transmission of nerve signals associated with sexual roles, occupation and hierarchical position in industrial culture.
The outlines are the “inside information of a person as a system”.
4 circuit can not exist without 3, 3 without 2, 2 without 1 ...
Certain instincts, behavioral programs, reflexes, emotions, ideological attitudes are contained in these contour categories.
In accordance with these circuits, human behavior is built in certain situations ... “Lack of intelligence” in computers takes place because they have no analogue to the first and other human circuits. Computers at the moment only tools. Perhaps artificial intelligence can be realized in a similar human system, for example, 4 contour.
One of the possible options for creating an AI is as follows:
Stage 1. Creation of a platform, with some internal information, on the basis of which it is possible to produce a training system.
The platform should model the human consciousness, with functions: memory, logic, associative, neurological logic, realization of an external language (English) and internal (figurative thinking) *, a module of emotions (reinforcing failures and victories - if unsuccessful, the reluctance to repeat is fixed, and if it is successful, memorization ). There should also be a certain “self-programming” module - the “curiosity function”.
Stage 2 Learning platform on various situations.
* - Translation of thoughts into words can be represented as database operations. Those. images are stored in one table, in the process of thinking with them are carried out operations of comparison, synthesis, etc., the results are stored in this or another table. The words of the language are stored in another table - the words are associated with images (one word with many, one with one, many words and one way) and thoughts are set out in accordance with this table. Or all these actions occur in one multidimensional database. In the synthesis of "erroneous meta-images", such as a "round square", "glass bath", the protective function is turned on - laughter ...
When implementing a database of images, the principles of encoding images are not clear. Alternatively, it is possible to establish links between images, images and words, and permissible and unacceptable combinations of images and their connection with words.
Human thinking is organized with the help of neural circuits and circuits. All information can theoretically be presented in a database, with complex relationships between tables and cells, but only when using non-Aristotelian, and three-digit, fuzzy logic, Tsad's logic when linking images - data in the database. The principle of data organization will remain identical, the only difference is in the methods of representing the organization of links. Taking into account the “principle of relativity of information” , one can say that the “hypothesis about human thinking” is simply presented from two points of view: “neurological” and “database points”.
14*. However, there are some reasons for suggesting that a person is more than his body, brain, neural connections, neuro-contours, etc. A person can control his body very effectively, like Yoga, for example, to change his world view - that is, reprogram some properties of your neuro-circuits, effectively manage your memory, etc. The question is who does this? After all, it is quite problematic to “pull yourself by the hair out of the swamp”?
Perhaps this is the metaprogramming contour (7th) ...
“The outline of metaprogramming — known as“ soul ”in Gnosticism,“ no-mind ”(y-blue) in China, White Light of Emptiness in Tibetan Buddhism, shiva-darshan in Hinduism, the True Intellectual Center of Gurdjieff — is simply the brain, which awakened the consciousness of himself. An artist who sees himself in his picture, who sees himself who sees himself in his picture ... According to the Zen metaphor, this is a mirror that reflects everything, but is not tied to anything. This is a conscious mirror that knows that it can always reflect something else by changing the angle of reflection. ” R. Wilson, "The Psychology of Evolution."
Considering the non-prevalence of this phenomenon in most people, it seems very problematic to imprint this mechanism from AI, so it must be designed with this contour in mind.
I will give the analogy of the contour metaprogramming:
- “person” from clause 4 * of this text,
- “self-awareness” of paragraph 7 *,
- “curiosity” of paragraph 8 ...
From the point of view of the biological structure of the brain - the mind, consciousness is not some thing or a metaphysical object, but what the brain does (or rather, “neocortex”), i.e. process, the action that he performs in his life.
“Now I hope that I have convinced you that the mind is just a name for what the brain does. It is not a separate thing that manipulates or coexists with neurons of the brain. Neurons are just cells. There is no mystical force that causes specific nerve cells or groups of nerve cells to behave differently than they normally would. ” Jeff Hawkins, "On the Intelligence."
This statement rests on the theoretical model of the brain as a “memory - prediction” process.
When modeling AI based on the contour model, AI certainly does not need 1, 2, 4 contours for its own functioning, but they will be useful for adequate communication with people. With the presence of 7 contours of metaprogramming, the AI ​​will be able to change the “imprints” of 1,2,4 contours and communicate adequately with various people. In fact, computers already possess the third circuit (and “only them”), the question is in effective use.
Database design should take into account multi-structured data storage. For example, there is an absolute concept of a circle, each person has it. The concepts “wheel”, “ball projection”, “sun”, i.e. all "round". For example, wheels include such concepts as “wooden wheel” (carts), “tire with tire”, etc. Those. In the memory-prediction model of Jeff Hawkins, the organization of the data and their interrelationship, including the “interdependence”, play a big role.
It can also be assumed that the “multi-structure” of the surrounding processes, i.e. such a property of world perception, which invests structures into structures into structures, etc., i.e. what is illustrated by the following quote ...:
“Mathematician J.W. Dunn turned this question into a parable. The artist, who escaped from a lunatic asylum, into which he (rightly or unfairly) was hijacked, decided to paint a field where he found himself. When he finished, he studied the result and realized that something was missing, namely, his and his canvases, which are part of the field. He began to paint again and depicted himself and the canvas in the middle of the field. Returning to the result philosophically, he realized that something was still lacking: namely, his canvas, in which he portrays himself and the canvas in the middle of the field. He began to redraw the picture a third time ... then the fourth ... and so on ad infinitum. ” R. Wilson, "The Psychology of Evolution."
... is a consequence of the structure and functioning of the neocortex, which uses patterns to perceive and store information about the world around ...
All incoming data in the brain, go in the form of patterns in the neocortex, where there are already certain structures (Liri contours) that represent the worldview of a single person.
(at the moment, the modeling of the metaprogramming module looks problematic, but this is more likely a question of developing Artificial Intelligence, and for Artificial Intelligence :) - approx. the author
nineteen*. Some properties of information models.
Transformations (some result) performed in one information model can be transferred to another model after a corresponding adjustment (i.e., “fitting” to the model to which the results are transferred) *.
* - This allows the “fact” that an information model describes one objective reality (which is subjectively an information model like a meta-model).
An information meta-model is a model that allows for the existence of other information models and allows for various transformations with them and in them and transferring the results obtained in one information model to another.
An analogue of the information model for human consciousness and the brain is a reality tunnel in NLP and a world view in a general sense.
to be continued ...
12*. From the concept of Alan Turing, “if the interlocutor cannot determine whether he is talking to a person or a machine, this proves that the machine behaves as smart as a person”, it follows that if a PC with a free conversation with a person yourself as a person, then the PC has intelligence.
Not every person can talk to any person on any topic. For example, not every programmer-specialist in the field of any language will be able to support a conversation with a scientist biologist on the topic of “reproduction of ciliates of shoes”. In everyday life, people often cannot find common topics for conversation ... etc.
')
Any person in the workplace is currently little different from a “computer”, i.e. from a computer with a specialized program to perform certain functions. For example, when a non-standard task appears, employees who do not have information about the methods of its solution, as a rule, pass it on to other employees, at a level higher or lower. The program also gives an error message. That is, for a person who orders a ticket via the Internet and buys it at the airport’s ticket office, the difference between a person and a computer is insignificant, and if you add certain functions to the program, or select a specific cashier at the airport, the person can no longer distinguish the person from the program ...
Also, “conversation” can be interpreted as communication, the transmission of information which is owned by the interlocutors.
But this includes the following aspects.
Suppose there are several categories of information. And each subsequent category of information is nested in the previous one, and in order to process it, you must have access to top-level information.
Example:
1. Section.
Category.
Article.
Those. to read the article, you need to get access to the category, to get access to the category, you need to get access to the section.
Take also the "outlines of consciousness" Leary - Wilson.
1. Biovietal (unicellular) consciousness. Characterized by the perception, processing and transmission of nerve signals responsible for cellular health and reproduction. Metabolic safety.
2. Emotional-motor consciousness (consciousness of mammals). It is characterized by the perception, processing and transmission of neuromuscular signals responsible for the mobility of the body and control over the territory. (Fight or run?)
3. Symbolic-artifact consciousness. Perception, integration and transmission of information using gestures and abstract symbols (first alphabet). Creating language and tools. Symbolic thinking.
4. Industrial consciousness. Perception, integration and transmission of nerve signals associated with sexual roles, occupation and hierarchical position in industrial culture.
The outlines are the “inside information of a person as a system”.
4 circuit can not exist without 3, 3 without 2, 2 without 1 ...
Certain instincts, behavioral programs, reflexes, emotions, ideological attitudes are contained in these contour categories.
In accordance with these circuits, human behavior is built in certain situations ... “Lack of intelligence” in computers takes place because they have no analogue to the first and other human circuits. Computers at the moment only tools. Perhaps artificial intelligence can be realized in a similar human system, for example, 4 contour.
One of the possible options for creating an AI is as follows:
Stage 1. Creation of a platform, with some internal information, on the basis of which it is possible to produce a training system.
The platform should model the human consciousness, with functions: memory, logic, associative, neurological logic, realization of an external language (English) and internal (figurative thinking) *, a module of emotions (reinforcing failures and victories - if unsuccessful, the reluctance to repeat is fixed, and if it is successful, memorization ). There should also be a certain “self-programming” module - the “curiosity function”.
Stage 2 Learning platform on various situations.
* - Translation of thoughts into words can be represented as database operations. Those. images are stored in one table, in the process of thinking with them are carried out operations of comparison, synthesis, etc., the results are stored in this or another table. The words of the language are stored in another table - the words are associated with images (one word with many, one with one, many words and one way) and thoughts are set out in accordance with this table. Or all these actions occur in one multidimensional database. In the synthesis of "erroneous meta-images", such as a "round square", "glass bath", the protective function is turned on - laughter ...
When implementing a database of images, the principles of encoding images are not clear. Alternatively, it is possible to establish links between images, images and words, and permissible and unacceptable combinations of images and their connection with words.
Human thinking is organized with the help of neural circuits and circuits. All information can theoretically be presented in a database, with complex relationships between tables and cells, but only when using non-Aristotelian, and three-digit, fuzzy logic, Tsad's logic when linking images - data in the database. The principle of data organization will remain identical, the only difference is in the methods of representing the organization of links. Taking into account the “principle of relativity of information” , one can say that the “hypothesis about human thinking” is simply presented from two points of view: “neurological” and “database points”.
14*. However, there are some reasons for suggesting that a person is more than his body, brain, neural connections, neuro-contours, etc. A person can control his body very effectively, like Yoga, for example, to change his world view - that is, reprogram some properties of your neuro-circuits, effectively manage your memory, etc. The question is who does this? After all, it is quite problematic to “pull yourself by the hair out of the swamp”?
Perhaps this is the metaprogramming contour (7th) ...
“The outline of metaprogramming — known as“ soul ”in Gnosticism,“ no-mind ”(y-blue) in China, White Light of Emptiness in Tibetan Buddhism, shiva-darshan in Hinduism, the True Intellectual Center of Gurdjieff — is simply the brain, which awakened the consciousness of himself. An artist who sees himself in his picture, who sees himself who sees himself in his picture ... According to the Zen metaphor, this is a mirror that reflects everything, but is not tied to anything. This is a conscious mirror that knows that it can always reflect something else by changing the angle of reflection. ” R. Wilson, "The Psychology of Evolution."
Considering the non-prevalence of this phenomenon in most people, it seems very problematic to imprint this mechanism from AI, so it must be designed with this contour in mind.
I will give the analogy of the contour metaprogramming:
- “person” from clause 4 * of this text,
- “self-awareness” of paragraph 7 *,
- “curiosity” of paragraph 8 ...
From the point of view of the biological structure of the brain - the mind, consciousness is not some thing or a metaphysical object, but what the brain does (or rather, “neocortex”), i.e. process, the action that he performs in his life.
“Now I hope that I have convinced you that the mind is just a name for what the brain does. It is not a separate thing that manipulates or coexists with neurons of the brain. Neurons are just cells. There is no mystical force that causes specific nerve cells or groups of nerve cells to behave differently than they normally would. ” Jeff Hawkins, "On the Intelligence."
This statement rests on the theoretical model of the brain as a “memory - prediction” process.
When modeling AI based on the contour model, AI certainly does not need 1, 2, 4 contours for its own functioning, but they will be useful for adequate communication with people. With the presence of 7 contours of metaprogramming, the AI ​​will be able to change the “imprints” of 1,2,4 contours and communicate adequately with various people. In fact, computers already possess the third circuit (and “only them”), the question is in effective use.
Database design should take into account multi-structured data storage. For example, there is an absolute concept of a circle, each person has it. The concepts “wheel”, “ball projection”, “sun”, i.e. all "round". For example, wheels include such concepts as “wooden wheel” (carts), “tire with tire”, etc. Those. In the memory-prediction model of Jeff Hawkins, the organization of the data and their interrelationship, including the “interdependence”, play a big role.
It can also be assumed that the “multi-structure” of the surrounding processes, i.e. such a property of world perception, which invests structures into structures into structures, etc., i.e. what is illustrated by the following quote ...:
“Mathematician J.W. Dunn turned this question into a parable. The artist, who escaped from a lunatic asylum, into which he (rightly or unfairly) was hijacked, decided to paint a field where he found himself. When he finished, he studied the result and realized that something was missing, namely, his and his canvases, which are part of the field. He began to paint again and depicted himself and the canvas in the middle of the field. Returning to the result philosophically, he realized that something was still lacking: namely, his canvas, in which he portrays himself and the canvas in the middle of the field. He began to redraw the picture a third time ... then the fourth ... and so on ad infinitum. ” R. Wilson, "The Psychology of Evolution."
... is a consequence of the structure and functioning of the neocortex, which uses patterns to perceive and store information about the world around ...
All incoming data in the brain, go in the form of patterns in the neocortex, where there are already certain structures (Liri contours) that represent the worldview of a single person.
(at the moment, the modeling of the metaprogramming module looks problematic, but this is more likely a question of developing Artificial Intelligence, and for Artificial Intelligence :) - approx. the author
nineteen*. Some properties of information models.
Transformations (some result) performed in one information model can be transferred to another model after a corresponding adjustment (i.e., “fitting” to the model to which the results are transferred) *.
* - This allows the “fact” that an information model describes one objective reality (which is subjectively an information model like a meta-model).
An information meta-model is a model that allows for the existence of other information models and allows for various transformations with them and in them and transferring the results obtained in one information model to another.
An analogue of the information model for human consciousness and the brain is a reality tunnel in NLP and a world view in a general sense.
to be continued ...
Source: https://habr.com/ru/post/46829/
All Articles