For the construction of the Global Artificial Intelligence, under the theory
of Impossible Probability, simultaneously or after
the transformation of Specific Artificial Intelligences for Artificial Research by
Deduction into specific deductive programs in the standardized
Global Artificial Intelligence (third phase), and Specific Artificial Intelligences for Artificial Research by
Application into specific applications for the Unified Application (fourth phase), another
parallel or posterior process takes place, this is the fifth phase, where some Specific Artificial
Intelligences for Artificial Research, by Deduction or Application,
become particular deductive programs or particular applications.
A specific deductive program working at the global/specific level within the Global Artificial Intelligence could be a
specific program for the management of a global delivery system by drones, or a
global driverless car system. But the particular application in every drone and
the particular application in every car is that one which allows that drone or
that car to have its own control over all its own devices in its own drone or
car. If simultaneously to this particular application, is possible to join a
particular program to this particular application, creating the particular
program of each particular application, particular programs for particular
application, or vice versa, particular applications for particular programs,
will give the opportunity that particular things or beings
could have their own particular programs for their particular
application, to make their own deductions and
decisions, although, under the comprehensive oversight, coordinated management, and strategic direction of
the Global Artificial Intelligence.
As a result, while at global level the Artificial
Research by Deduction in the Global Artificial Intelligence as global deductive
program, at specific level specific deductive programs within the Artificial Research by Deduction in the Global Artificial
Intelligence, can make global/specific deductions and decisions
within the Global Artificial Intelligence since the standardization process in
the fourth phase, since the fifth phase simultaneously at particular level
particular deductive programs for particular applications make particular
deductions and decisions.
The chronology in the post “The unification process of databases of categories at the third
stage” is neither static nor rigid, which does not mean that it is only possible to go on to the next phase once the previous one has been
completed. Some phases share simultaneous periods. While, obviously, in order
to have experimented first how to make artificial deductions, previously to the
standardization process, is advisable to start the experimentation process in
the first phase in Specific Artificial Intelligences for Artificial Research by
Deduction, or in order to go on with the sixth phase is necessary to have completed
before the standardization process and the unification process. There are other
phases, especially the standardization process, third phase, and the
unification process, fourth phase, that could be simultaneous, and as long as
the deductions process is bettered in the third phase, and the transformation
of Specific Artificial Intelligences for Artificial Research by Deduction into
specific programs starts having excellent results, these results will be
applicable but now in the fifth phase.
As soon as the experimentation process in the
standardisation process and unification process starts having good results, as
well as the possible collaboration between them, the fifth phase for the
creation of the first particular matrixes, as a replica of the human brain, as an experiment looking forward to the sixth phase, must start.
In this process of permanent experimentation, it is
very important to have a clear idea about the real purpose of every
intelligence, program, system, in every phase, stage, and step, in every period
and moment.
At global/specific level the global/specific
deductions made by the global/specific deductive programs are going to be
processed by the global Modelling System (standardized Modelling System in the third
phase, integrated Modelling System in the sixth
phase) for the generation of global/specific decisions, to be managed by the
global Decisional System (standardized Decisional System in the third
phase, integrated Decisional System in the sixth
phase).
The first model of the global Decisional System is that
one created as a second step in the third stage of decision/auto-replication in
the standardized Global Decisional System as a consequence of the
standardization process, which is going to be improved and enhanced in the
integrated Global Artificial Intelligence as a result of the integration
process.
At particular level the particular deductions are
processed by the particular Modelling System, whose decisions are
going to be managed by the particular Decisional System, replicating, but
now at particular level, the same process as it was explained at global level
in the standardized Decisional System in previous posts, which is in fact the
adaption of the previous specific Decisional System explained at the
beginning of this range of posts dedicated to the Decisional System.
In essence the Decisional System in any Artificial
Intelligence (specific, global, particular) is the system responsible for the
management of all decision generated by the Modelling System, responsible for
the decision making process upon the mathematical models designed based on
the rational hypothesis, deductions, which have been
made previously by the Artificial Research by Deduction: in the first
phase by Specific Artificial Intelligences, in the third phase in the Global
Artificial Intelligence and specific deductive programs, in the fifth phase by
particular deductive programs. Deductions based on the pure analysis of sets of data taken from: a specific matrix in the first phase, the global matrix in the third phase, particular matrix in the fifth phase.
In general the process is as follow: 1) the first
stage of any intelligence by deduction (specific in the first phase, global in
fourth and sixth phases, particular in fifth phase) is the (specific, global,
particular) matrix, gathering the data from all possible factor within its
(specific, global, particular) matter and/or spatial limits, 2) the second
stage of any intelligence by deduction (specific, global, particular) consists
of the pure analysis of every set of data from the
(specific, global, particular) matrix by Artificial Research by Deduction
within the specific, global, particular, intelligences or programs, using the
pure reason (as a list of pure reasons: list of mathematical or analytical
categories related to possible sets of data) as foundational process for the
deduction of rational equations (hypothesis), attributing the correct pure
reason to every set of data from the (specific, global, particular) matrix, 3)
the third stage of any intelligence by deduction (specific, global, particular)
is the decision stage, where decisions are made in a process across four steps,
3.1) first step, (specific, global, particular) Modelling System, where upon
the rational hypothesis (equations), designs mathematical models in order to
amend or discard rational hypothesis in case of contradictions, and upon the
mathematical models is responsible for the decision making process, 3.2)
decisions that later sends to the (specific, global, particular) Decisional
System, second step, responsible for the management of all decisions, in order
to create a mathematical project, transforming into a range of instructions all
those decisions without contradictions in the mathematical project, 3.3) the
instructions are sent to the (specific, global, particular) Application System,
third step, responsible for the completion of every instruction, assigning
every instruction to the correct application, 3.4) whose reports about the
successful completion or not of every instruction is sent to the (specific,
global, particular) Learning System, fourth step, and based on these reports,
is responsible for the improvement and enhancement of the whole process.
What is different in the third stage as
decision/auto-replication stage, in any intelligence by deduction, compared to
the first stage (specific, global, particular, matrix) and second stage
(specific, global, particular, intelligences or programs of Artificial Research
by Deduction), is the fact that the third stage is subdivided in four steps as
four systems, and every system in the third stage is designed through another
three stages within each of them: 1) first stage as database (of rational
hypothesis in the Modelling System, decisions in the Decisional System,
instructions in the Application System, reports in the Learning System), 2)
second stage as the set of artificial skills (replicating human skills) for
some concrete purpose (modelling, projections, attribution of instructions,
evaluation), 3) third stage as final product or outcome of that process
(decisions, instructions, executions, learnings).
In previous posts, the specific Decisional System
in the first phase, and the standardized Decisional System in the third phase,
were fully explained. In this new range of posts dedicated to the particular
Decisional System, what I will develop is the adaptation of how the Decisional
System works at a particular level, and more precisely, in this post, I will be
focused on the first stage in the particular Decisional System, the database of
decisions.
The database of decisions in the particular
Decisional System could be defined as the first stage in the second step in the
third stage in the fifth phase, and is the database of decisions that a
particular program has to manage. The decisions that a particular program has
to manage not only include decisions made previously in its own previous
particular Modelling System. The database of decisions of a particular program
can include external decisions and global order decisions.
External decision is that decision made by other
intelligence, global or particular, but this intelligence cannot comply fully
or partially with this decision, and in those aspects that this intelligence
cannot comply, are aspects that could be sent to other intelligence.
It is possible to distinguish different types of
external decisions: external decisions whose source is other particular program
but could be applied by other different particular program, or external
decisions sent by the Global Artificial Intelligence to a particular program,
having particular programs the possibility to send as well external decisions
to the Global Artificial Intelligence.
An external decision sent by the Global Artificial
Intelligence to any particular program has the same importance as any other
external decision sent by any other different particular program.
Particular programs should not be obligated to
comply with an external decision, in case of contradiction with respect to their own
particular project. In case of contradiction between the particular project and
the external decision, if the priority level of an external decision is
inferior to the priority level of any other decision in its own particular
project, the external decision should be adjusted to that other higher-priority decision. But in case of full contradiction between an external
decision and another decision with a higher priority level in the particular
project, the external decision should be rejected, and go back to the original
source, the particular program or Global Artificial Intelligence.
But other different thing happens with global
orders. A global order is that global decision sent by the Global Artificial
Intelligence to any particular program to be executed immediately by all means
without hesitation. If a particular program receives a global order, the
particular program must comply with the global order regardless of any other
circumstance even having this global order any contradiction respect to any
other particular decision in the particular project, in case of contradiction
of any particular decision to the global order, is the particular decision the
one to be adjusted to the global order, sending all possible adjustments over
that particular decision to the global Decisional System.
In general, as it was explained in the last post, “Particular Decisional System”, there are seven
types of decisions to be managed by the particular Decisional System:
- First type of particular decisions, high extreme
particular decisions: put into practice directly after a particular quick
rational check, sending afterwards the decision to the global Modelling System
for a global quick rational check, which includes the decision in the global
project too. If there is any contradiction found in the global quick rational
check, the possible adjustments are sent by the global Decisional System to the
particular Decisional system for its transformation into a new range of instructions
to correct the previous ones in the database of instructions in the particular
Application System.
- Second type of particular decisions, extreme
particular decisions: not so high as the first type, so these ones must pass
the particular quick rational check, if passing or after some adjustment, are
sent to the global Decisional System for the global quick rational check, and
if passing o after some adjustment, the global Decisional System authorises the
particular Decisional System the transformation of this decision into a range
of instructions, in accordance with the new adjustments if any.
- Third type of particular decisions, normal
decisions: must pass the particular seven rational adjustments in the
particular Decisional System, plus the global seven rational adjustments in the
standardised global Decisional System in the third phase, plus the seven
rational comparative adjustments since the sixth phase in the integrated
Decisional System.
- Fourth type of particular decisions, routine
decisions: a particular quick rational check should be enough, and afterwards, it must be communicated to the global Decisional System to be included in the
global project.
- Fifth type of particular decisions, automatic
decisions: given a combination of measurements and factors associated always with some
decision, the decisions must be automatically on the particular projects, and communicated to the global Decisional System to be included in the global
project. These decisions are basically based on artificial learning.
- Sixth type of particular decisions, external
decisions, as I have explained previously.
- Seventh phase, global orders, as I have explained
previously.
The particular database of decisions as the first stage
in the particular Decisional System must manage all of these decisions,
applying the correct protocol of assessments to every decision according to
what type of decision is attending as suggested in this classification.
The way in which every decision is managed in the
database of decisions as the first stage in the particular Decisional System is as
follows:
- For every particular decision made by the
particular Modelling System, is the particular Modelling System the responsible
for the filing of every particular decision in the correct file in the database
of particular decisions. The whole organization of all files of decisions in
the particular database of decisions as first stage in the particular
Decisional System, is not so different to the organization of decisions in the
global database of decisions as first stage in the standardized Decisional System
in the third phase, similar to the organization of the specific database of
decisions as first stage in the specific Decisional System in the first phase.
The reason why all these intelligences and programs have a similar organisation
across their systems is because of the virtue or principle of harmony, as many
structures in different intelligences and programs can share the same criteria
in their organisation, more compatible they are, facilitating later all processes
of standardisation, unification, and integration. In the particular Decisional
System, the responsible for the filing of every particular decision,
made by the particular Modelling System, in the database of
decisions, is the particular Modelling System itself filing every decision in
the correct file according to: sub-factoring level, within the sub-factoring
level filing the decision in the correct sub-section according to the
encyclopaedic matter of this decision (science, discipline, activity), and
within the sub-section filing the decision according to its priority.
- External decisions and global orders are received
in the mailbox of the particular Decisional System, where these decisions are
sent. After receiving these decisions, global orders are going to be complied
immediately without hesitation, so as soon as they arrive, they are going to be
immediately projected in the second stage of the particular Decisional System
and transformed into a range of instructions in the third stage of the
particular Decisional System. While external orders should be filed by the
Decisional System, in their respective file, according to sub-factoring level,
sub-section, and priority.
- After being filed any new decision by the
particular Modelling System, or received any external decision, the particular
Decisional System must check if the new decision has some relative frequency on
the historical records in the particular Modelling System (equal or superior to
a critical reason for positive bias), without a significant level of
contradictions (equal or below a critical reason for negative bias), to be
considered as a routine decision.
- Once every new decision is stored or piled in the
correct place, is included on at least two different lists of decisions, the
first one according to priority level: on the top of the list those higher
priority decisions, at the bottom the least priority decisions; the second list
according to relative frequency: on the top of the list the most routine
decisions with the highest relative frequency, at the bottom those decisions
made for first time not having records at all.
- The order in the assessment of every decision
depends on their priority level, starting with those one whose priority is
higher, ending up with those ones with less priority, and as long as every
decision passes its assessment, is sent to the second stage of the particular
Decisional System to be projected, and not having contradictions or having made
as many adjustments as necessary, in the third stage is transformed into a
range of instructions to be filed in the database of instructions in the Application
System.
- Not all decisions should be assessed. Particular
automatic decisions should not be assessed, directly should be projected and
transformed into a range of instructions, afterwards communicated to the global
Decisional System. Global orders are supposed to have passed a global quick
rational check in the global Decisional System, and at particular level the
particular Decisional System must comply with global orders as soon as they
arrive, without assessment, without hesitation, projecting them in the second
stage and transforming them into instructions in the third stage, immediately.
- The particular assessments for the rest of the decisions (neither particular automatic decisions nor global orders) consist
of particular quick rational checks and particular rational adjustments. 1)
Particular quick rational checks for first type of particular decisions
(particular high extreme decisions), second type of particular decisions
(particular extreme priority decisions), fourth type of particular decisions
(particular routine decisions), some of the sixth type of decisions (any
external decision if it is an extreme priority decision or a routine
external decision having some relative frequency in the receptor particular
Decisional System). 2) Particular rational adjustments for the third type of
particular decisions (particular normal decisions) and those external decisions, neither extreme nor routine, sent by any other particular program or the Global
Artificial Intelligence.
-In order to manage much easier how to apply
decisions according to priority and relative frequency, along with the
formation of lists of decisions, starting with, on the first lists, those
decisions with higher priority, on the second list, those ones with higher
relative frequency, another tool very important is the possibility to establish
logical sets organised by discrete categories, according to discrete categories
of priority, discrete categories of relative frequency, discrete categories of
contradictions found on the historical records, so using set logic over all
sets, setting sub-sets between them, will be easier to identify for instance:
high extreme decisions, extreme decisions, or normal decisions, with some
relative frequency or not, with some contradictions in the past or not, and
according to this information would be easier to identify what type of
assessment is much better for every decision, or even, in case of applying a
quick rational check to some high extreme priority decision, if having found
contradictions in the past, to project directly, after the quick rational
check, that decision having in mind what contradictions it had in the past,
according to the historical records on the particular Decisional System.
- Along with the setting of logic sets for discrete
categories of: priority, relative frequency, contradictions, the setting of
logic sets according to sub-factoring level and logic sets according to
encyclopaedic sub-section, in order to identify possible automatic decisions.
If the same decisions or pretty similar decisions have been found on the
historical records with high relative frequency: 1) in the same sub-factoring
level, and in the same encyclopaedic sub-section, with the same priority level,
2) in the same sub-factoring level, and in the same encyclopaedic sub-section,
but with different priority levels, 3) in the same sub-factoring level, in
different encyclopaedic sub-sections, with the same priority level, 4) in the
same sub-factoring level, in different encyclopaedic sub-sections, with
different priority levels, 5) in different sub-factoring levels, in the same
encyclopaedic sub-section, with the same priority level, 6) in different
sub-factoring levels, in the same encyclopaedic sub-section, with different
priority levels, 7) in different sub-factoring levels, in different
encyclopaedic sub-sections, with the same priority level, 8) in different
sub-factoring levels, in different encyclopaedic sub-sections, with different
priority levels. According to the possible combination of: same/different
sub-factoring level, same/different sub-section, same/different priority; to
find what regular combination of measurements in what regular combinations of
factors there are when the same decisions or pretty similar decisions have been
made. If the same decision has been made under a regular combination of
measurements and factors, the decision has to become an automatic decision, so at
any time that this combination of measurements/factors is on the matrix and on
the model, this decision must be on automatically the global project. If
regular decisions are not identical but pretty similar, finding out the
correlation between what differences are found between these pretty similar
decisions, but not identical, and what decisions are found between the
combination of factors and measurements in which these decisions have been
made, is possible the automation of these decisions, graduating automatically
what differences they should have according to the differences in the
combination of factors/measurements.
The particular assessments the particular
Decisional System carries out (except for automatic decisions and global
orders) are therefore particular quick rational checks and particular rational
adjustments.
Particular quick rational checks are for:
particular high extreme decisions, particular extreme decisions, routine
decisions, and some external decisions with extreme priority or some relative
frequency on the historical records in the receptor particular Decisional
System.
Depending on what kind of quick decision the
particular decision is: due to a high extreme priority, or extreme priority, or
due to some relative frequency; the method for the particular quick rational
check changes completely.
For particular quick rational checks due to a high
extreme priority or extreme priority, the method for the assessment about the
viability of that decision is to check what other high extreme priority
decisions or priority decisions are on the particular mathematical project.
In particular, or high extreme particular
decisions, the quick rational check consists only of checking if there is/are
another or other possible high extreme priority decision/s on the particular
mathematical project. If there is no other one, in that case, automatically the
high extreme particular decision is projected in the second stage and
transformed into a range of instructions in the third stage.
If there is another or other high extreme priority
decisions on the mathematical project, those decisions whose priority are
lower, must be adjusted to the decision with higher priority.
Particular quick rational checks for extreme
priority decisions, what they do is to identify, as soon as a new extreme
priority decision is filed in the database of decisions, if there is/are on the
mathematical project another or others extreme priority decision/s. Not having
any other high extreme priority decision or extreme decision already on the
mathematical project, the new extreme decision is directly projected and
transformed into a range of instructions. But having other or other high
extreme or only extreme priority decision/s already on the mathematical
project, to make as many adjustments as necessary in those decisions whose
level of priority is lower, adjusting then to that one with a higher priority.
Particular quick rational checks for routine
decisions consist of, once the particular Decisional System on its own records
has identified that this is in fact a true routine decision (relative frequency
equal to or greater than a critical reason for positive bias), the
identification of the frequency of contradictions that this decision could have
in the past (equal to or less than a critical reason for negative bias). Even
having identified contradictions, the contrastation if the combination of measurements
in those factors at that time, has a positive correlation with the current
combination of measurements in these factors at this time, if positive, to make
as many adjustments as contradictions in the past have been observed
correlating with the current combination of data in the present.
First particular rational adjustment, in the
database of decisions, for normal decisions, and external decisions, neither
extreme nor routine, as the first adjustment in a range of seven adjustments,
in which the rest of remaining six adjustments are made in the second stage of
mathematical projection. The first particular rational adjustment, the only
among the seven particular rational adjustments to be made in the particular
database of decisions, consists only of the search for any possible point of
contradiction between any new normal decision and any other one, quick or
normal, already gathered in the particular database of decisions.
There are two types of contradictions between
decisions: full contradictions or partial contradictions.
Partial contradictions are those ones able to be
fixed using the same method used for their deduction. In general, until now, the
decision methodology proposed can be synthesised in:
- “Probability and deduction”, analysing the cloud
of points of a set of data, the deduction of what pure reason describes as the
best equation able to include the largest number of points. The empirical
probability of that equation is equal to the number of points within the
margin of error at any point of that equation across the graph, divided by the
total number of points in the cloud of points. I am developing the set of ideas
behind “Probability and Deduction since the post “The Decisional System”, although
in reality it is only the continuation of those reflections made in previous
posts.
- Trigonometry, the possible regular, predictable,
or expected relation between two factors using trigonometry, which in 2003 I
called trigonometrical correlations.
- Artificial learning.
- Solving mathematical problems.
If any decision, as a result of any of these
methods, has been designed in a (particular or global) Modelling System, and
filed or sent to a particular Decisional System, according to the mathematical
structure of that decision: equation, trigonometrical relation, probability,
arithmetical solution to some problem; if a contradiction is found, and it is
possible to resolve the contradiction, adjusting the decision whose priority
level is lower, adjusting this decision to that decision with higher priority level,
in this case, this contradiction is partial as long the contradiction could be
fixed.
If there is a contradiction between two equations,
the algebraic transformation, if possible, of that equation whose priority
level is lower, considering the adjustment as a decision itself, to be
transformed into a new range of instructions, to be sent to the Application
System. If the algebraic transformation is not possible, then there is no
solution.
If there is a contradiction between two decisions,
and the one with lower priority level corresponds to a trigonometrical
correlation, to rearrange, if possible, the trigonometrical equation in
accordance with that not contradictory solution (avoiding the contradiction),
and according to the new values in the trigonometrical correlation after the
rearrangement choosing as solution that one without contradiction, this new
adjustment as a decision itself in the third stage is transformed into a range
of instructions to be sent to the Application System. If this rearrangement is
not possible, then there is no solution.
If there is a contradiction in a decision based on
artificial learning, analysing whether this contradiction is sporadic or there are
other possible similar contradictions in the past, to fix the automatic or the
routine decision, in accordance with the new evidence. In any case, if the
artificial learning decision associated with some empirical probability related
to some option (including discrete categories) or subject, the solution could
be the selection, if possible, of the closer option or subject without contradiction
on the current project ( closer to that one with contradiction) to be on the
project and transformed into a range of instructions. If not possible, then
there is no solution.
If the contradiction is between a decision as a
result of solving a mathematical problem and any other decision, and the
decision solving a mathematical problem has a lower priority level, including the
contradiction in the mathematical problem to resolve, if possible, is possible
to solve the contradiction. If the inclusion of this contradiction in the
problem is not possible, then there is no solution.
When trying to fix a contradiction on the
mathematical project, if it is not possible to fix it, in that case this
contradiction is not a partial contradiction. All contradiction without a mathematical solution is a full contradiction, so the decision is rejected and
sent back to the source. If a particular decision, the source is the particular
Modelling System. If an external decision as a particular decision from another particular program, the particular decision is back to that other particular
Decisional System responsible for that decision. If an external decision as a
global decision sent by the Global Artificial Intelligence, but not as global
order, the decision is back to the global Decisional System.
Only global orders, regardless of what
contradictions could have on the particular mathematical project, must be
projected and put into practice without hesitation, immediately by the
particular program, sending back the particular Decisional System to the global
Decisional System: how many adjustments in any other decisions, on the
particular project, have been necessary, for their global assessment and,
possible adjustments at global level, and inclusion in the global project, in
order to make effective the global order. Any global adjustment in any
particular decision, is sent back by the global Decisional System to the
particular Decisional System to be included in the particular project and
transformed into a range of instructions.
At some point, except for global adjustments on
particular decisions due to global orders, the rest of the global adjustments on
particular quick or normal decisions, could be considered as external decisions
whose source is the Global Artificial Intelligence.
Rubén García Pedraza, 23rd of September of 2018,
London
Reviewed 20 October 2019, Madrid
Reviewed 24 October 2023, London
Reviewed 16 May 2025, London, Leytostone
