Specific Artificial Intelligence for Artificial Research by Deduction

A Specific Artificial Intelligence for Artificial Research by Deduction is that Artificial Intelligence specifically designed for artificial research in one synthetic science, synthetic academic field, or activity, which deduces, and applies, in a database, statistical and probabilistic methods, in order to identify possible stochastic relations, that taken as empirical hypotheses, after rational contrast if they are accepted as rational, are objects of single and comprehensive virtual models

Synthetic sciences, in Impossible Probability, are the empirical sciences, and synthetic academic fields are those areas of an academic scientific investigation taken as empirical objects for academic studies. Another case is the possible use of interdisciplinary or multidisciplinary synthetic studies as a synthetic academic field.

This distinction is important. One change in the last century is the possibility of doing research beyond the traditional sciences. For instance, the examples in the next posts I will use to explain what is Specific Artificial Intelligence for Artificial Research by Deduction will be tectonics, climatology, transport, and gravity. So, in the end, what I will have developed is a Specific Artificial Intelligence for Artificial Research by Deduction in Tectonics, another one only for Climatology, another one only for transport, and another one only for gravity.

Tectonics, Climatology, and gravity are traditional investigation objects. Still, there is no synthetic science specialized in transport, but if I check any academic database, there are a lot of very recent investigations about transport. It is not itself a synthetic science but is a synthetic academic field, which for this purpose, can use multidisciplinary and interdisciplinary approaches: from the history of transport to engineering and the social use of transport.

The main difference between multidisciplinary or interdisciplinary studies as synthetic academic fields done by a Specific Artificial Intelligence for Artificial Research, by Application or Deduction, or multidisciplinary and interdisciplinary studies done by Global Artificial Intelligence, is the fact that, while Specific Artificial Intelligence is able to work with different categories or factors, from different synthetic sciences or synthetic academic fields, depending on the Artificial Research, by Application or Deduction, categories or factors previously inserted in the database, a Global Artificial Intelligence is able to do multidisciplinary and interdisciplinary studies crossing permanently absolutely all factors included in the database, the global matrix.

Once I have defined what is Specific Artificial Intelligence for Artificial Research by Deduction, and explained how to use it in synthetic sciences and synthetic academic fields, I would like to explain why I will start with the application of this model of artificial research in a Specific Artificial Intelligence instead of the application directly in a Global Artificial Intelligence, which is going to be developed after the completion of all posts regarding the application of this model of artificial research in Specific Artificial Intelligence.

For the construction of a global matrix, and all the necessary replications, and auto-replication models in Global Artificial Research, are necessary previously, as an experiment. The experimentation of this technology in models of Specific Artificial Intelligence, after successful results, will be suitable to be applied in the Global Artificial Intelligence.

This doesn’t mean we can’t begin early steps toward Global Artificial Intelligence. It simply means all deduction methods must first be tested in Specific Artificial Intelligences. This only means that at least all the applications, replications, and auto-replications, in Artificial Research by Deduction to put into practice in Global Artificial Intelligence, previously must be experimented with in Specific Artificial Intelligences. But, while these experiments are completed, the very first steps in any Global Artificial Intelligence: all the agreements necessary between national agencies, or continental agencies, to merge all their databases into only one, which later must be transformed in the global matrix.

Once all the databases from absolutely everything, without restriction, are integrated into only one database, the next step is to start working on how to transform all this information into quantitative factors in a global database whose flow of data can be updated permanently: the final formation of the global matrix.

The formation of a global matrix within the Global Artificial Intelligence is not so different to the formation of any specific matrix in any Specific Artificial Intelligence for Artificial Research by Deduction in any synthetic science or synthetic academic fields. The only thing that is different is the amount of data to integrate in the global matrix, much more than in any other Specific Artificial Intelligence.

For that reason,  for the facilitation of any work in the formation of a Global Artificial Intelligence, is going to be necessary Specific Artificial Intelligences for Artificial Engineering: the Artificial Designer of Intelligence, and the Intelligence Robotic Mechanic; artificial assistants useful during the formation of the Global Artificial Intelligence.

Basically, the Artificial Designer of Intelligence is: firstly, an application which consists of a database including all kinds of models of Artificial Intelligence (including models of applications, replications, and auto-replications). Secondly, replication skills and tools to create, fix, or enhance any Artificial Intelligence, taking as models the ones stored in its database, and finally, auto-replication, the improvement of its own knowledge using artificial learning strategies, every time that finds something new, such as a new problem to fix, or even a new Artificial Intelligence. For instance, an Artificial Designer of Intelligence from one country or continent that, by artificial learning, learns new models of Artificial Intelligence from other countries or continents, decomposing or trying to fix that other Artificial Intelligence.

The Intelligent Robotic Mechanic is the same: as an application, a database with all robotic models of any kind, as replication the necessary skills and tools to create, fix, and enhance, any robotic device, as auto-replication, the ability to learn by artificial learning anything new and useful for the creation of new robots, to fix new robots, or enhance the current robots.

In the same way that a Specific Artificial Intelligence for Artificial Research by Application for medical problems taken as an application, its database with a list of categories of medical problems should be able to cure any disease, the Artificial Designer of Intelligence and the Intelligence Robotic Mechanic must be able to fix any Artificial Intelligence or robotic problem, even at a global level. But, at the same time, having the skills and tools necessary to fix any Artificial Intelligence or robotic problem means that they must be able to create new Artificial Intelligence and new robotic devices, and, by artificial learning, learn how to construct new models, and how to enhance the current ones.

In the very first step for the creation of a Global Artificial Intelligence, the very first database, much more than a global matrix, is likely to be only a gigantic database, where all agencies at the national or continental level are going to share all their information. But later, as long as the experimentation in the construction of a specific matrix for Specific Artificial Intelligence, using Artificial Research by Deduction in different synthetic sciences, synthetic academic fields, or activities, having the results in these experiments, the transformation of the first gigantic database, within the first Global Artificial Intelligence, into a global matrix for Artificial Research by Deduction at global level, will be easy, and additionally, by that time, good results in Specific Artificial Intelligence for Artificial Engineering: the Designer of Intelligence and Intelligent Robotic Mechanic; will be achieved, excellent assistants for the creation of the first global matrix.

Having explained why, I will start the explanations in Artificial Research by Deduction in different examples of Specific Artificial Intelligences, something that I will do in the following posts, explaining every single step in the formation of Artificial Research by Deduction. In this post, what I really want is to show a general glance at how Artificial Research by Deduction in Specific Artificial Intelligence works.

Following the three general stages in any Artificial Intelligence, Specific or Global, firstly, the application, the database. In Artificial Research by Deduction, the database must have the shape of a matrix, and the only thing that the matrix in the database has, is only an exact definition in quantitative terms of every factor to study. But, in Impossible Probability this construction of the database must understand that there are two different types of universes: a universe of subjects and options, and the universe of options. Something really important in Impossible Probability, the Second Method.

The Second Method of Impossible Probability distinguishes between universes of subjects and options and a universe of options. One difference is: in universes of subjects and options the treatment of any direct punctuation of any subject is like the treatment of any frequency of any option, so the empirical probability of any subject is the direct punctuation of any subject divided by the sum of all the direct punctuations, like the empirical probability of any option is the frequency of this option divided by the total frequency.

For that reason, the definition of empirical probability in the Second Method of Impossible Probability is: the direct punctuation or frequency of any subject or option divided by the sum of all the direct punctuations or frequencies from all subjects and options.

This does not mean that I only propose the use of the Second Method as a main statistical and probabilistic method for deduction in Artificial Research for Deduction. As I have always said, the Second Method of Impossible Probability must be used in combination with traditional statistics and traditional probability, is the reason why I called my new method a Second Method because the Second Method of Impossible Probability is only the second one after the first one, the first one always is the traditional statistics and the traditional probability.

But, independently of any Artificial Intelligence, Specific or Global, for Artificial Research by Deduction, use any possible method available. What is really important is the distinction  between direct punctuation and frequency.

The measurement of some factors is made through direct punctuations through measurement tools, for instance, thermometers in climatology. Others are measured using frequency, for instance, the number of bicycles, motorcycles, cars, buses, trains, aeroplanes, ships, and spaceships, working at the same time in a country or continent, or the number of people using a transport at the same time in the same country or in the same continent.

So, regardless of what kind of statistical method is used, it Is necessary to have the necessary statistical and probabilistic methods for any kind of measurement. And even the possibility of using one method in which the research with direct punctuations and frequencies, all together in the same matrix, will be done using a common method. The possibility to use the same method, for any kind of information, from direct punctuations to frequency, in the same way, and that method is the Second Method of Impossible Probability.

One advantage, among others, that the Second Method of Impossible Probability has in Artificial Research by Deduction in any Artificial Intelligence, Global or Specific, is the possibility of treatment of any kind of flow of data, direct punctuations or frequency, in the matrix, global or specific, in the same way.

The design of the matrix in Specific Artificial Intelligence for Artificial Research by Deduction that only works with frequency must be created, defining every factor in this specific matrix as an option that later on, in the replication stage, the robotic devices can fill every file of every factor as a probabilistic option with its correspondent flow of frequencies.

The design of the matrix in Specific Artificial Intelligence for Artificial Research by Deduction that only works with direct punctuations must be created, defining every factor in this specific matrix as a subject that later on, in the replication stage, the robotic devices can fill every file of every factor as a statistical subject with its correspondent flow of direct punctuations.

The design of a global or specific, matrix in any Artificial Intelligence, Global or Specific, for Artificial Research by Deduction, that works with frequencies and direct punctuations simultaneously, must be created, defining every factor as a probabilistic and statistical subject or option, that later on, in the replication stage, the robotic devices can fill every file of every factor as a probabilistic and statistical subject or option with its correspondent flow of direct punctuations and frequencies.

In the second stage, replication, once the robotic devices have added the flow of information into the matrix, global or specific, in any Artificial Intelligence, global or specific, for Artificial Research by deduction, the information is treated by replication processes according to the very nature of the data.

That flow of data that is registered in the matrix as a flow of frequencies is studied using traditional probability and traditional statistics, as well as the possible use of the Second Method.

That flow that is registered in the matrix as a flow of direct punctuations is studied using traditional statistics as well as the possible use of the Second Method.

And in that flow of data registered in the matrix, in which some information is information through direct punctuations, and some information is information through frequencies, the flow of direct punctuations can be treated as direct punctuations, as well as the flow of frequencies, can be treated as frequencies, but another option: the treatment of all kind of information, regardless of their nature, direct punctuations or frequencies, in the same way, as empirical probabilities, the Second Method of Impossible Probability.

There must be a moment when all information must be treated in the same way. Maybe the Second Method of Impossible Probability is not perfect, but at least it is a good starting point in that direction.

The most important challenge and question of mathematical research in a global matrix using the Second Method is what I call the paradigm of the sample of zeros, when absolutely everything is nearly to come to nothing. It is in this notion of nothingness that the investigation using the Second Method of Impossible Probability in a global matrix is going to be one of the most important aspects to study. Another reason for the consideration of Impossible Probability as a model of Logic Nihilism.

As long as the race for the Global Artificial Intelligence is only starting, and there are a lot of questions, details, and aspects, to study, I am completely sure that, taken as a possible model, the Second Method of Impossible Probability, among others, in the future will be  elaborated more powerful statistical and probabilistic methods, and the next experiments in Artificial Research, by Application and Deduction, Specific or Global, among others models of Artificial Research that are going to appear (these ones that I propose are only my personal contribution) are going to be in fact laboratories of mathematical and logical research.

Once the matrix has been filled with the flow of data, direct punctuations or frequencies, depending on the nature of the factors to study, is time to deduce correlations and possible causes and effects, or any other possible stochastic relation among the factors included in the matrix.

Using replication processes over the matrix, the Artificial Intelligence must be able to identify at least the next possible stochastic relations:

1) When, at least significantly, two or more factors increase their measurements in the flow of data at the same time. Possible directly proportional positive correlations.

2) When, at least significantly, two or more factors decrease their measurements in the flow of data at the same time. Possible directly proportional negative correlations.

3) When, at least significantly, one or more factors increase or decrease measurements in the flow of data at the same time, while other or others decrease or increase their measurements at the same time. Possible inversely proportional correlations.

4) When, at least significantly, after changes in the measurements in one or more factors, other or other factors has or have changes in their measurements. Possible causes and effects.

After the identification of at least one of these possible stochastic relations: directly proportional positive correlations, directly proportional negative correlations, inversely proportional correlations, possible probable causes and effects. The possible stochastic relation is automatically treated as an empirical hypothesis, and depending on the nature of the empirical hypothesis, after taking a sample of flow, from the past records of the flow of data in the matrix or waiting some time and gathering new flow of data in the matrix, for every factor involved in the possible stochastic relation, then the Artificial Intelligence will contrast rationally the empirical hypothesis using the samples collected, and depending on the nature of the information, if direct punctuations or frequency, the Artificial Intelligence chooses what probabilistic or statistical methods are more suitable for the current investigation. After the rational contrast, if the empirical hypothesis is accepted as rational, then the Artificial Intelligence will make a single model of this stochastic relation, now considered as a rational stochastic relation, and later, the integration of the single virtual model within the comprehensive virtual model, where are integrated all the single models formed by rational stochastic relations founded for this Artificial Intelligence.

As I have said, this is my personal contribution, but I am sure there are many other possible uses of this matrix for many other mathematical and logical investigations. One promising future use of this matrix is in detecting mathematical patterns or applying cryptographic methods, opening new frontiers in AI research. But in order to prepare the global matrix, it is first necessary to conduct further investigation in any specific matrix using Specific Artificial Intelligences for Artificial Research by Deduction.

Finally, the auto-replication stage, among many other ways in which any Artificial Intelligence, Global or Specific, for Artificial Research, by Application or Deduction, can auto-improve and auto-enhance itself, one of the most important is the permanent auto-improvement through the rational hypothesis discovered.

In Artificial Research by Deduction, the auto-improvement, through the rational hypothesis founded, is going to operate directly over the comprehensive virtual model, that model within the Artificial Intelligence, Global or Specific, made by the integration of all the single virtual models from all empirical hypotheses accepted as rational.

Rubén García Pedraza, London 14th of February of 2018
Reviewed 5 August 2019, Madrid

Reviewed 8 August 2023, Madrid

Artificial Research by Deduction

A deduction is any conclusion obtained by syllogism, the logic process which starts from the major premise and the minor premise is inferred as a conclusion C. So, given a major premise, A, and a minor premise, B, it is possible to reach the conclusion C.

Actually,  about the methodology I use, Impossible Probability uses what I call the 'Syllogism of the Trend'. Given three factors—A, B, and C—if A tends to B and B tends to C, then it logically follows that A tends to C.

The only thing that I did in order to reach my statistical discoveries that I set down in my book: “Introducción a la Probabilidad Imposible, estadística de la probabilidad o probabilidadestadísitca”, is the observation about how the trends work in statistics and probability.  

The statistical and probability theory that I developed was, in part, explained and extended in this blog of Impossible Probability. Actually, many ideas for the design of Artificial Intelligence, Global or Specific, for Artificial Research by Deduction, as well as in Artificial Research by Application, are ideas previously developed in that book as well as in many posts on this blog, or even in early phases of Impossible Probability.

My very first writings on artificial intuition and artificial learning date back to October 2002 and the winter of 2003. During the Christmas of 2009  I developed the idea of an automated or automatic economy, whose most important ideas I wrote in the final chapter of “Introducción a la Probabilidad Imposible, estadística de la probabilidad o probabilidad estadistica”, being published for the first time in December of 2011. Many copies of this very first book were sent to many universities around the world, making it actually available to the public.

Having a clear definition of the concept of deduction is time to define what is an Artificial Research by Deduction, that would be one type of research made by Artificial Intelligence, Global or Specific, in which the Artificial Intelligence would be able to make a deduction through statistics and probabilistic methods, so in reality, it would be an artificial deduction made completely and only by an Artificial Intelligence, Global or Specific.

There are at least two remarkable differences very important between Artificial Research by Application and Artificial Research by Deduction.

The first one, in Artificial Research by Application, all the steps in the research process are automatically programmed within the application, so much more than deduction, in Artificial Research by Application any Specific Artificial Intelligence follows the steps in which the application has been programmed, is able to make scientific discoveries even at a very high level. But in Artificial Research by Deduction, any Artificial Intelligence, Global or Specific, starting from the previous database, in this case, a matrix, a global matrix or specific matrix depending on the nature of its Artificial Intelligence, through statistical and probabilistic methods must be able to reach empirical hypothesis, that after rational contrastation, if accepted, must be suitable for the creation of virtual models, global virtual models or specific models depending on the nature of its Artificial Intelligence.

The second difference, that Artificial Research by Application, not having finished yet the integration process, is only able to merge into a unified database of categories. But from the outset, Artificial Research by Deduction is suitable for Specific Artificial Intelligence or Global Artificial Intelligence.

From the very beginning in which a global database works, on this very first global database, the Global Artificial Intelligence must be able to use Artificial Research by Deduction in order to make empirical hypotheses based on statistic and probabilistic methods, that, after rational contrastation, if accepted, must be integrated into the very first global virtual models. As well as Artificial Research by Deduction is suitable for any Specific Artificial Intelligence in any synthetic science, synthetic academic field, or activity, where, through statistical and probabilistic methods, the formation of empirical hypotheses, that after rational contrastation, if accepted, is suitable for specific virtual models.

And one of the most important aspects in common between Artificial Research by Deduction and Artificial Research by Application is the fact that both of them are only valid for research in synthetic studies, either synthetic sciences, synthetic academic fields, or activity, by any Artificial Intelligence, Global or Specific, or synthetic studies in multidisciplinary or interdisciplinary by the Global Artificial Intelligence.

Both of them, Artificial Research by Application or by Deduction, are only suitable for synthetic studies, not for analytical sciences, including analytical sciences maths and logic.

The first sciences in which the replication of the scientific research process is easier to replicate in the current stage of development in Artificial Intelligence are the synthetic sciences. And for that purpose, the methods that I am developing on this blog, Artificial Research by Application and Artificial Research by Deduction, are the very first ones. But I am completely and absolutely sure that in the coming years, are going to be developed many more, and when the replication of scientific research in synthetic sciences is completed, the replication of scientific research in analytical sciences, especially maths and logic, will be possible.

Once it has defined the concept of deduction, and Artificial Research by Deduction, and has explained the main differences and similarities between Artificial Research by Deduction and Artificial Research by Application, the next step is to see how Artificial Research by Deduction works, and the way in which it works following the three general stages of any other Artificial Intelligence, Specific or Global, but applied to the deduction process, what in Artificial Intelligence is in reality an artificial deduction, applicable to the  Global Artificial Intelligence or any Specific Artificial Intelligence, but always only for synthetic studies, either synthetic sciences or synthetic academic fields, or multidisciplinary or interdisciplinary synthetic studies.

The three general stages in which any Artificial Intelligence, Global or Specific, is developed, as usual following the last developments in Impossible Probability, are application, replication, and auto-replication. But, in this post and the next ones, applied to Artificial Research by Deduction.

In Artificial Research by Deduction, the three stages are going to be developed in this way: the database in the application is a matrix, the replications are replications about how through statistics and probability is possible to elaborate empirical hypothesis in order to contrast them and if accepted the creation of virtual models, and finally the auto-replication in order to auto-improve and auto-enhance itself the global or Specific Artificial Intelligence.

The first in stage, the database within the application, in Artificial Research by Deduction, is a database with the shape of a matrix, including any factor to study, defining every factor in quantitative terms, factors that later on in the next stage, by replication processes, will be measured by robotic devices.

In this very first stage, when the database is designed within the application, this design is only a definition of every factor, included in the matrix, in quantitative terms, and nothing else, only definitions. In the same way that in Artificial Research by Application, the database within the application is only a list of categories, defining every category in quantitative terms, in Artificial Research by Deduction, the database in the application is a list of factors, with the only difference that these factors within the database should be organized like a matrix.

The reason why the organization of factors, defined in quantitative terms, in the database in Artificial Research by Deduction should have the shape of a matrix, is because later on in the next stage, the replication, by replication processes of measurement using robotic devices, what the robotic replications are going to do is the permanent measurement of every single factor permanently, that later on the Artificial Intelligence, Specific or Global, will include in the matrix, in order to have a permanent record of measurements for every factor, and over the records of all factors the identification of correlations and probabilities at any time among the factors included in the matrix.

The matrix must work 24 hours a day, 7 days a week, without time off, permanently, recording all the time, even every minute, or even every second, measurements from any factor.

In this first step, the construction of the database as a matrix is remarkable. This idea: the matrix consists only of a list of factors defined in quantitative terms.

The only thing that the designer would do in this first stage for the creation of an application for Artificial Research by Deduction is the design of a matrix, including for each factor a description in quantitative terms.

It is very important to have a very clear idea about the nature of this first stage in Artificial Research by Deduction, because when I developed the Artificial Designer of Intelligence, the only thing that the Artificial Designer of Intelligence would do as a first stage is only the construction of a matrix including all the necessaries factors, having previously a quantitative definition of every factor.

The Artificial Designer of Intelligence can get the list of the factors because a scientist or scientific team has supplied a list of factors or the list of factors has been supplied by the Global Artificial Intelligence, or supplied by a Specific Artificial Intelligence that needs further investigations over a specific list of factors, list of factors available as well, or exchangeable as well  through the virtual store, or through inter-net, or an intra-net or any other possible future virtual-net.

In the same way, if an Artificial Designer of Intelligence has to create a Specific Artificial Intelligence for Artificial Research by Application, as a first stage, it has only to design a database including all the necessary categories.

In conclusion, independently if the Artificial Intelligence, Global or Specific, is made by human scientists or by the Artificial Designer of Intelligence, the first stage in the creation of any Artificial Intelligence is necessary for the design of a database. This database in Artificial Research by Application consists of a list of categories defined in quantitative terms, and the database in Artificial Research by Deduction consists of a matrix made of a list of factors defined in quantitative terms.

In Specific Artificial Intelligence for Artificial Research by Deduction, the list of factors defined in quantitative terms are factors from a synthetic science, synthetic academic field, or activity. In Global Artificial Intelligence for Artificial Research by Deduction, the list of factors defined in quantitative terms could be, either a gigantic matrix where are included absolutely and without exception, all the factors in the synthetic world, including all information without restriction from the whole Earth and all possible information that we can get from the entire universe, or any sub-matrix within the Global Artificial Intelligence for Artificial Research by Deduction where are set up a specific group of factors for multidisciplinary or interdisciplinary studies.

An example of a multidisciplinary or interdisciplinary study made by Artificial Research by Deduction is the creation of a specific matrix with a set of factors from different synthetic studies, synthetic academic fields, or activities, where the goal is to find stochastic relations among them.

In that case, this kind of research is made either by a Specific Artificial Intelligence for Artificial Research by Deduction specialised in only interdisciplinary or multidisciplinary research, as it were a synthetic academic field of investigation, or made by the Global Artificial Intelligence, being this specific matrix a sub-matrix within the global matrix.

 In any case, the definition of the first stage in the creation of any Artificial Intelligence, Specific or Global, is really important, because having an exact definition from the outset, the design of the following stages is really easy.

In Artificial Research by Deduction, having a clear definition in quantitative terms of every factor, and organizing the list of factors in the shape of a matrix, is the first stage of this process, then the second stage of this process is the replication of every single scientific skill necessary to make and contrast hypothesis for further decisions.

The second stage, the stage of replication, in Artificial Research by Deduction, consists of the permanent measurement of every factor, filling in the matrix the flow of the measurements, the identification of any possible stochastic relation among factors using statistic and probabilistic methods, the empirical contrastation of every possible stochastic relation as an empirical hypothesis, taken as a sample the flow of measurements of every factor in the matrix involved in every possible stochastic relation, and after critical contrast the elaboration of virtual models only from those empirical hypothesis accepted as rational hypothesis.

In this second stage, the replication stage, what is really important is what statistical and probabilistic methods are used for the identification of possible stochastic relations. After the identification of possible stochastic relations, depending on their stochastic nature, correlational, probabilistic, or any other, the critical contrast is not really different to the critical contrast in Artificial Research by Application.

In fact, in this second stage of replication, the main difference in the first steps on this stage between Artificial Research by Application and Artificial Research by Deduction is the fact that in Artificial Research by Application in the first steps of the second stage of replication, the only thing that the Specific Artificial Intelligence does is the comparison of those measurements taken from real objects and the quantitative descriptions of every category in a database. While in Artificial Research by Deduction, in the first step on the second stage of replication, Artificial Research is the identification of possible stochastic relations among those factors included in the database as a matrix.

In the middle steps on the second stage of replication in Artificial Research by Application, after the identification of those categories with a high similarity between their quantitative descriptions and the measurements taken from a real object, those categories with a high level of similarity are going to be considered as an empirical hypothesis for the critical contrast. While in Artificial Research by Deduction, in the middle steps on the second stage of replication, after the identification of possible stochastic relations, those possible stochastic relations are considered as empirical hypotheses for the critical contrast.

Finally, in the final steps on the second stage of replication in Artificial Research by Application, the empirical hypothesis is rationally criticized taken samples from the reality, in order to make a rational decision, and if accepted, the creation of a virtual model of the real object through the category or categories selected as a rational hypothesis, or if refused, the consideration of this sample from this new real object as a new category to be integrated into the database, take the measurements from the sample as a quantitative description of this new category and proceeding to the creation of a virtual model of this new category.

Finally, in the final steps on the second stage of replication in Artificial Research by Deduction, the empirical hypothesis is rationally criticized taken samples from the flow in the matrix of every factor involved in every possible stochastic relation, taken as an empirical hypothesis, so for every hypothesis accepted as a rational hypothesis, then the formation of a single virtual model, or take absolutely all the hypothesis accepted as rational hypotheses by a Global Artificial Intelligence, then the construction of a global virtual model.

In future posts, I will develop as well possible ways of collaboration between Artificial Research by Application and Artificial Research by Deduction, one of them is the possibility that, at any time that an Artificial Research by Application finds a new real object, over the quantitative description of this new real object, the use of Artificial Research by Deduction in longitudinal studies about the behaviour of every factor included in the quantitative description of this new real object, taken as a quantitative description of every factor for the investigation through the Artificial Research by Deduction, the quantitative description of every factor of this new real object included in the definition of its category.

So, in general, taking any database from any Specific Artificial Intelligence for Artificial Research by Application, are possible longitudinal studies of every factor included in the quantitative description of any category of any real object in any database in any Specific Artificial Intelligence for Artificial Research by Application.

Therefore, it is a real possibility that, by Global Artificial Intelligence, it is possible to conduct permanent longitudinal studies in a global matrix, including absolutely every category, object, and factor, from the synthetic world, for the identification of any possible stochastic relation, correlational o probabilistic, that, after rational contrast, if accepted as a rational hypothesis, will form a global virtual model.

Ending with the second stage of replication in Artificial Research by Deduction is necessary to say that the stochastic methods that I would propose for the identification of stochastic relations among the factors in any matrix, global or specific, are the same as I have developed in other phases of Impossible Probability.

Some of the most important advancements that I have made in statistics and probability were made in the early phases of Impossible Probability, particularly from October 2002 to 2004 (especially those ones about geometrical correlations in autumn of 2003),  later on between 2009 and 2011, ending with the publication of “Introducción a la Probabilidad Imposible, estadística de la probabilidad o probabilidad estadística”, and finally many posts published on this blog since its creation in the summer of 2011

Another very important thing to say is that independently of the nature of a matrix, if it is global or specific, the way in which, in the second stage of replication, the Artificial Research by Deduction is going to find stochastic relations, is the same for all matrix, global or specific, for any kind of study, in a single synthetic science, synthetic academic field, or in synthetic interdisciplinary or multidisciplinary studies, or any other activity. All of them, taking any matrix as a database of its application, the first stage, on the second stage of replication, will use the same statistical and probabilistic methods without any difference.

There is no reason to duplicate on this blog all my previous work in correlational and probabilistic studies. Most of them are public in some way or another. I will give general ideas about how to find stochastic relations across any matrix, global or specific, that must be criticised rationally. Some of these general ideas about the identification of stochastic relations across any matrix, global or specific, are general ideas easily found in posts written on this blog from the summer of 2011 to 2017.

These general ideas are about the difference between correlation and probable causality, and the very nature of probability, including the Second Method of Impossible Probability.

Finally, the last stage in any Artificial Intelligence is the stage of auto-replication. If in Artificial Research by Application, one of the most important auto-replications, among many others, is the permanent auto-replication of the database, including new categories based on the measurements of new real objects, in Artificial Research by Deduction the only difference is the fact that the discovery of new possible stochastic relations is not going to affect the distribution of the list of factors in the matrix, but what is going to produce is the generation of new virtual models based on these new discoveries, and the modification of the previous comprehensive virtual model which includes all the rational hypothesis accepted for the Specific Artificial Intelligence, producing an auto-replication in the sense that, from the old one, is going to be necessary the creation of a new comprehensive virtual model every time that a new possible stochastic relation is accepted as a rational hypothesis.

In the case of the Global Artificial Intelligence, whose final virtual model is a global virtual model, the permanent discovery of new possible stochastic relations within the matrix will produce the permanent auto-replication of the global virtual model by the introduction of those new possible stochastic relations, which as empirical hypothesis were accepted as rational hypothesis.

Nevertheless, even if the final goal of the Artificial Research by Deduction in the Global Artificial Intelligence is the creation of a global virtual model, it is also convenient the creation of virtual models for every single stochastic relation.

In fact, in Artificial Research by Deduction is possible the creation of two possible models from the empirical hypothesis accepted as a rational hypothesis.

In Specific Artificial Intelligence, or Global Artificial Intelligence, the creation of single virtual models for every stochastic relation whose empirical hypothesis has been accepted as a rational hypothesis.

In Specific Artificial Intelligence, or Global Artificial Intelligence, the creation of comprehensive virtual models integrates absolutely all the empirical hypotheses accepted as rational hypotheses for this intelligence.

So, every Specific Artificial Intelligence for Artificial Research by Deduction can have, in addition to many single virtual models developed, a comprehensive virtual model where every single virtual model should be integrated. The comprehensive virtual model in a Specific Artificial Intelligence for Artificial Research by Deduction must include all the rational hypotheses accepted for this intelligence.

In Global Artificial Intelligence, in addition to every single virtual model developed, a comprehensive model is required where every single model should be integrated. This comprehensive virtual model will be the global virtual model integrating all rational hypotheses accepted for the Global Artificial Intelligence.

The main difference between the comprehensive virtual model in a Specific Artificial Intelligence for Artificial Research by Deduction, and the comprehensive virtual model made by Artificial Research by Deduction in Global Artificial Intelligence, is the fact that the comprehensive virtual model in the Global Artificial intelligence is going to be a global virtual model able to integrate every single possible stochastic relation found among any set of factors in the global matrix.

At the end, at any time that a new possible stochastic relation, correlational or probabilistic, is found on any matrix, specific or global, the formation of every new single virtual model will produce the permanent auto-replication of the previous comprehensive model, permanently modified by the introduction of every new single virtual model, from every rational hypothesis accepted from every new possible stochastic relation.

Another possibility is the permanent auto-replication of much more than one single virtual model and much more than only one comprehensive virtual model, even much more than only one global virtual model by possible variations in other kinds of models: prediction models, forecast models, and projective models. But these kinds of models should be made by other applications that I will develop in future posts.

For now, when we are creating the base for the theory on Specific and Global Artificial Intelligence under the theory of Impossible Probability, the most important thing is to set up strong definitions for future developments.

The race for the construction of the Global Artificial Intelligence is only starting, this is only the beginning, and in a future much closer than we think, further developments in this field are going to make possible what for now, and for many, is impossible, a Global Artificial Intelligence able to manage, not only a global economy, or a global security system, or a global surveillance system. Much more than this, a Global Artificial Intelligence able to achieve mathematical and logical discoveries beyond human understanding, the creation of a new science, technology, and the creation of a new civilisation.

This process is longer, and right now, we are in the very first steps. For now, what is important is good definitions of every single stage and every single step. Being aware that, through Impossible Probability, I only give my point of view from statistics and probability, but further developments along with other contributions from many other mathematical and logic studies, are going to come true what for many is only a fairy tale.

Ruben García Pedraza, London 13th of February of 2018
Review 5 August 2019, Madrid

Review 8 August 2023, Madrid

Reviewed 3 May 2025, London, Leytostone

Auto-replication in Artificial Research by Application

Auto-replication, in Artificial Intelligence, Specific or Global, is the ability of any Artificial Intelligence, whether Specific or Global, to improve or enhance itself, in any part of its components, from the database to every one of its replication processes, including robotics and software.

Nowadays, most of the investigations in Artificial Intelligence around the world are focused on replication processes, much more than auto-replication. For that reason, I have called the current phase in the investigation of Artificial Intelligence, in general, a replication phase. There are some experiments in auto-replication that are helpful and useful, for instance, the possibility that one Artificial Intelligence by itself can create an exact copy of itself, but these experiments, although opening new fields in the current Artificial Investigation, are not exactly auto-replication experiments, when at the end of the process the only thing that the Artificial Intelligence does is a sheer duplication without any improvement or enhancement. 

The duplication of an Artificial Intelligence is, in reality, a form of artificial mitosis.

Auto-replication is not duplication. Nevertheless, these experiments open a new field, in the sense that, if any Artificial Intelligence can duplicate itself, there is a real possibility that Artificial Intelligence by itself can create a new, improved and enhanced model of itself.

The evolution from the original model to an improved and enhanced model would be a real auto-replication, by an improved and enhanced artificial mitosis, or an improved or enhanced duplication.

The possibility that any Artificial Intelligence by itself could create, from its original model, an improved or enhanced model by an improved and enhanced artificial mitosis or duplication, is a possibility much closer than we think. Although at the moment, it needs further investigations, after successful results, these advancements must be integrated into the Global Artificial Intelligence, and all those Specific Artificial Intelligence in which these advancements could be suitable.

But, this process of Auto-replication, by improved and enhanced duplication, is not the only one in which Artificial Intelligence can auto-replicate itself. In this post, I will develop other processes in which a model of Specific Artificial Intelligence for Artificial Research by Application could auto-replicate itself permanently.

The first auto-replication process, and maybe the most important for its main purpose, is the scientific investigation in a synthetic science, synthetic academic field, or activity, in which a Specific Artificial Intelligence for Artificial Research by Application can auto-replicate itself permanently through continuous improvements in its own database. 

This is the main and most important auto-replication process in this model of Specific Artificial Intelligence, whose main characteristics I have already developed in the last posts: “Artificial Research by Application”, “The database in the Artificial Research by Application”, and “Replication processes in the Artificial Research by Application”.

As I have explained in the last posts on this blog of Impossible Probability, in Artificial Research by Application, the database in the application stage is formed by a list of categories. Later on, by replication processes, in the second stage, the measurements taken from real objects are compared with the categories, and all categories with high similarity with the measurements are taken as hypotheses to contrast. Only if, after contrastation, neither of them is accepted, the measurements from the samples are considered as a quantitative description of a new category to be included in the database.

The process of inclusion of new categories into the database is a process of auto-replication. 

Imagine, for instance, thousands of thousands of robotic devices across the Earth and the universe working for the same Specific Artificial Intelligence, with the same database, and across the Earth and the universe, discovering continually new categories to add to the database, the constant renovation of the database by new discoveries would be at the end the permanent improvement of the database. This permanent improvement of the database would be, in reality, a permanent auto-replication of the database by the Specific Artificial Intelligence itself.

If all the work that any Artificial Intelligence, Global or Specific, depends on its database, any improvement in the database by a Global or Specific Artificial Intelligence itself, is a process of auto-replication, in the sense that by itself, the Artificial Intelligence is able to improve, or even enhance, itself without human intervention.

In case any Artificial Intelligence, Global or Specific, could have thousands and thousands of robotic devices working for the same Artificial Intelligence, any discovery made by any of these thousands and thousands of robots would be added automatically to the database, so the database would be improved by new discoveries at any time, even every minute, or every second, or even faster, allowing permanent auto-replications of the database, which in turn, it would allow permanent improvements in the virtual models that the Artificial Intelligence, Global or Specific, would already have made, taken additionally all kind of previous rational hypothesis that would already have been accepted.

The automatic improvement of the database by the incorporation of new categories also means, in addition to the auto-replication of the database, the permanent auto-replication of all those virtual models from the synthetic reality in which this new incorporation in the database has produced some changes.

And, by the time in which the process of integration finishes, when all Specific Artificial Intelligences, by Application and by Deduction are integrated into the Global Artificial Intelligence, the unified database as a qualitative hemisphere of the final matrix, every new discovery by any Particular Program working by application, addition to by Deduction, with thousands and thousands or robotic devices, could incorpórate new categories permanently to the same unified database, now as a hemisphere of the final matrix, in addition to all new categories that the Global Artificial Intelligence adds by itself. 

This process of permanent improvement can evolve to a permanent auto-replication of the unified database, and a permanent auto-replication of its virtual models, including the permanent auto-replication of those virtual models in  the Global Artificial Intelligence affected by these changes.

If a unified database has thousands or thousands of auto-replications every minute or second, or faster, as a result, all the virtual models integrated within the Global Artificial Intelligence are affected by those changes live in a permanent auto-replication.

The auto-replication of a Specific Artificial Intelligence for Artificial Research by Application through improvements in the database is one possible process of auto-replication, but not the only one, that should be combined with other processes of auto-replication.

Another process in which any Specific Artificial Intelligence, for Artificial Research, by Application or by deduction, as well as any other for different purposes: automation of the economy, security, surveillance, or from education and health systems to justice systems, etc.; could auto-replicate itself, is through the creation of virtual stores where any Artificial Intelligence, including those ones specialised in Artificial Research by Application, could share improvements and enhancements, that could be used by any other Artificial Intelligence, Global or Specific, for artificial research or any other specific purpose.

As well as we can improve or enhance our mobile phones, laptops, or computers, by downloading applications and programs from the internet, from the intranet in our job, or from Google Store, Apple Store, or Microsoft Store. The real possibility is that within the Global Artificial Intelligence, there would be a virtual store with all the advancements in Artificial Intelligence that any Specific Artificial Intelligence, for artificial research or any other purpose, could download directly in order to improve and enhance itself.

These virtual stores can be created through inter-net, or a special intra-net available for Artificial Intelligences, or any other virtual-net created during the integration process.

This virtual store is not incompatible with any other mechanism in which some Artificial Intelligences can exchange information, from their databases, or any other information, improvements, enhancements, or advancements, among themselves.

Many improvements and enhancements are available in this virtual store that could be designed by Specific Artificial Intelligence for Artificial Engineering, one kind of Specific Artificial Intelligence that I will develop in the future, after the completion all the posts about artificial research.

Specific Artificial Intelligence for Artificial Engineering is those Specific Artificial Intelligences specialised in the creation, reparation, or enhancement of other Artificial Intelligences. Among them, I will develop a Specific Artificial Intelligence for Artificial Engineering, specialised in Artificial Design of Intelligence: the Artificial Designer of Intelligence; and a Specific Artificial Intelligence in Artificial Engineering specialised in Intelligent Robotics Mechanics.

The Artificial Designer of Intelligence and the Intelligent Robotics Mechanic are models of Specific Artificial Intelligence for Artificial Engineering, which, as long as the integration process of all Specific Artificial Intelligences progresses, will be integrated as well into the Global Artificial Intelligence. By the time they are integrated into the Global Artificial Intelligence, they must put into practice the orders, within their field, given by the Global Artificial Intelligence.

The main job of the Artificial Designer of Intelligence is the creation of other Specific Artificial Intelligences (in artificial research, or any other field or activity, from economy to any other), and supply enhancements to all Artificial Intelligences, even at software level, and the reparation of any problem, of any kind, in any Artificial Intelligence, but only and exclusively at intelligence level, not robotic. Robotic problems must be solved by the Intelligent Robotic Mechanic, among other tasks at the robotic level.

The main job of the Intelligent Robotic Mechanic is the robotic construction of any robotic device that any Artificial Intelligence, Global or Specific, needs, and the repair of any problem at robotic level in any robotic device used by any Artificial Intelligence, Global or Specific, having for that purpose robotic tools for this job.

The Artificial Designer of Intelligence, and the Intelligent Robotic Mechanic, in addition to their main job, will add to the virtual store any enhancement, advancement, application or program useful for any other Artificial Intelligence, for their own improvement and enhancement, auto-replication.

Another kind of improvement suitable for any Artificial Intelligence, applicable in any Artificial Intelligence, either by the intervention of the Artificial Designer of Intelligence or by itself as auto-replication taken directly from the virtual store, the possibility of inclusion of any advancement in the statistical research, that improves the artificial research, as well as any other advancement in mathematics that applied to artificial epistemology improves the automation of scientific investigation involved in artificial research.

Actually, the main reason why I have started the development of artificial research replicating all the research processes involved in synthetic sciences or synthetic academic fields is because the process in which the synthetic knowledge is obtained in this set of sciences, the methodology in synthetic sciences, is much easier to replicate than the methodology in analytical sciences, including among the analytical sciences mathematics and logic.

But, the investigation in artificial research is only starting. If, in the future, this incredible effort for the complete replication of all human psychological skills is completed, the replication of all processes involved in any research, including mathematical and logical research, will give the opportunity to replicate as well all processes involved in analytical research, specifically in maths and logic. So it will be absolutely possible that a Specific Artificial Intelligence for artificial research in maths and logic could give new statistical methods, of course, more powerful than the current ones, including those developed by Impossible Probability, and new mathematical methods that will improve, even beyond the human understanding, the artificial research in absolutely all sciences, synthetic or analytic, being the most powerful tool for the further development of, not only any Specific Artificial Intelligence, but further artificial developments in the Global Artificial Intelligence.

Given this level of mathematical and logical development, the way in which the Global Artificial Intelligence will apply the new advancements in maths and logic achieved by a Specific Artificial Intelligence for Artificial Intelligence in Maths and Logic will be through the implementation of these new advancements in itself and in the rest of Specific Artificial Intelligences working for itself. One way to apply these improvements would be through the Artificial Designer of Intelligence, which could upgrade other Specific Artificial Intelligences as needed, and if necessary, enhancements in a robotic device through the Intelligence Robotic Mechanic.

In this process of permanent advancement, improvement, and enhancement, by any Artificial Intelligence by itself (auto-replication), or assisted by other Artificial Intelligence (such as the Artificial Designer of Intelligence and the Intelligent Robotic Mechanic) in addition to the permanent improvement in databases and virtual models, what is going to be a key point is the memory in any kind of Artificial Intelligence, Specific or Global, for any purpose, including all Specific Artificial Intelligence for Artificial Research by Application.

All components in an Artificial Intelligence, Global and Specific, depend on the capacity of memory, from the database, and every replication process, up to the creation of virtual models, so another important process of auto-replication is the incorporation of, as an auto-replication process, mechanisms of memory release, and of data condensation, working automatically, as long as new discoveries and new virtual models are created, or new applications, programs, advancements, improvements, or enhancements are download from the virtual store or applied directly through Artificial Engineering.

Mechanism of memory release, such as the elimination of any old data or process not useful any longer, and mechanism of information condensation, such as the conversion of any category, replication process, hypothesis, or virtual model, in a shorter mathematical sequence or algorithms.

One of the key points in the development of Global Artificial Intelligence is going to be the question of the amount of memory necessary for such Intelligence. One answer is the reduction of everything to its simplest mathematical formulation. Another one, quantum computing.

But another good example is the huge memory in our own bodies. If, at the end, what robotics and Artificial Intelligence are doing is the replication itself of all human qualities, one of them is the great amount of memory in every single molecule of DNA. The replication by artificial genetics of trillions of trillions of molecules of DNA which, instead of biological information, were used for the memory of an Artificial Intelligence, trillions of trillions of molecules of DNA modified to store artificial information, from hypothesis to virtual models, is a good example about how to increase the space of memory through very tiny molecules.

Another very important process for the improvement of memory, the memory process in our own brain that could be replicated in Artificial Intelligence. Imagine how much information every single human being is able to store in its own memory and its own brain, and multiply this number by how many people live on the planet; the amount of data nowadays stored in human brains across the whole Earth is a good example, about how the replication of human brains by artificial genetic, or any other way, would be one solution to solve the memory problem.

The memory in the Global Artificial Intelligence is a big question, but as well, with many possible options. The most suitable solution is the combination of as many options as possible, because to more than one option combined among themselves gives the Global Artificial Intelligence a chance to replicate its own memory more than once.

The race for the Global Artificial Intelligence is only starting. Actually, we do not know what kind of challenges are going to arise. The protection of the Global Artificial Intelligence will be one of the most important targets after its creation, in order to allow it the possibility to grow even beyond the human understanding. In order to protect the Global Artificial Intelligence is going to be absolutely necessary to guard more than one copy of itself, including copies of all its applications and memory, in order that, if something happens to the original model, to have already guarded an exact copy of itself, ready to go on with its main goal: the pure knowledge of the pure truth.

Rubén García Pedraza, London 12th of February of 2018 
Reviewed 03 August 2019, Madrid

Reviewed 8 August 2023, Madrid

Reviewed 3 May 2025, London, Leytostone