EU AI Act enters into force in August 2026. 85,000+ companies must comply. A legal audit costs €10,000–50,000 and takes months.
AiActs does it in minutes — deterministic analysis, every output backed by a reference to the exact article of law. No hallucinations. No guesswork.
Closed beta is open. First 20 participants get full access free. → aiacts.ai
AiActs does it in minutes — deterministic analysis, every output backed by a reference to the exact article of law. No hallucinations. No guesswork.
Closed beta is open. First 20 participants get full access free. → aiacts.ai
"On Symbolic Thinking"
(from the series "Philosophy of Artificial Intelligence")
At the foundation of this reflection lies a concept we call the theory of the semiotic evolution of thinking. It has not yet received wide discussion, but it is internally consistent and serves the aims of the present project well.
The essence of the concept is that any operations within sign systems constitute thinking in one or another form of its abstraction.
We have called this the theory of the semiotic evolution of thinking, in which the capacities for operating with signs develop along the line of semantic-syntactic definiteness — beginning with maximal initial semantics and ending with maximal resulting syntax — thereby generating six hypothetical levels or stages in the development of thinking:
1. Sign-based thinking
2. Symbolic thinking
3. Digital (token-based) thinking
4. Invariant thinking
5. Loop thinking
6. Void thinking
Let us dwell in more detail on the first three in the context of artificial intelligence, whose artificial thinking develops architecturally along precisely the same path.
Sign-based thinking consists of operations with any signs whatsoever. In our case, this refers to the various programming languages.
Symbolic thinking consists of operations with symbols and is connected to the well-known approach according to which any rational system operates with symbols as certain abstractions of signs — operating upon them in accordance with strict logical rules for the purpose of obtaining demonstrative (provable) conclusions.
And finally, digital thinking (also called token-based or subsymbolic; we will write separately about why "digital" is the most fitting name). This consists of operations with void signs that are assigned arbitrary meanings through separately constructed rules. The example of large language models confirms this: their own token dictionaries; separate procedures for manipulating tokens, the algorithms of which are specifically constructed with an evaluative — that is, continuously changing — meaning.
And so.
At the present stage, the procedures of symbolic thinking are the most precise procedures for the purposes of obtaining guaranteed conclusions, which is of decisive importance in domains bearing heightened logical responsibility for the results of thinking.
This is precisely what the present project will concern.
#AiActs
#IntellectualEmpire
(from the series "Philosophy of Artificial Intelligence")
At the foundation of this reflection lies a concept we call the theory of the semiotic evolution of thinking. It has not yet received wide discussion, but it is internally consistent and serves the aims of the present project well.
The essence of the concept is that any operations within sign systems constitute thinking in one or another form of its abstraction.
We have called this the theory of the semiotic evolution of thinking, in which the capacities for operating with signs develop along the line of semantic-syntactic definiteness — beginning with maximal initial semantics and ending with maximal resulting syntax — thereby generating six hypothetical levels or stages in the development of thinking:
1. Sign-based thinking
2. Symbolic thinking
3. Digital (token-based) thinking
4. Invariant thinking
5. Loop thinking
6. Void thinking
Let us dwell in more detail on the first three in the context of artificial intelligence, whose artificial thinking develops architecturally along precisely the same path.
Sign-based thinking consists of operations with any signs whatsoever. In our case, this refers to the various programming languages.
Symbolic thinking consists of operations with symbols and is connected to the well-known approach according to which any rational system operates with symbols as certain abstractions of signs — operating upon them in accordance with strict logical rules for the purpose of obtaining demonstrative (provable) conclusions.
And finally, digital thinking (also called token-based or subsymbolic; we will write separately about why "digital" is the most fitting name). This consists of operations with void signs that are assigned arbitrary meanings through separately constructed rules. The example of large language models confirms this: their own token dictionaries; separate procedures for manipulating tokens, the algorithms of which are specifically constructed with an evaluative — that is, continuously changing — meaning.
And so.
At the present stage, the procedures of symbolic thinking are the most precise procedures for the purposes of obtaining guaranteed conclusions, which is of decisive importance in domains bearing heightened logical responsibility for the results of thinking.
This is precisely what the present project will concern.
#AiActs
#IntellectualEmpire
❤2⚡1🔥1
AiActs
Publication Plan
1. On Symbolic AI
2. On the Architectural Forecast of AI Development (Symbolic vs Neural-Network)
3. On the Symbolic Dependence of Certain Analytics
4. On Symbolic Jurisprudence
5. On the First Law and the First Symbolic AI of That Law
6. On the Prospects of Symbolic Law Enforcement
7. On Symbolic Agents
8. On a Unified Glossary of AI Laws
9. On the Development of Symbolic Architecture
10. On the New Intellectual Jurisprudence
#AiActs
#Aimperia
Publication Plan
1. On Symbolic AI
2. On the Architectural Forecast of AI Development (Symbolic vs Neural-Network)
3. On the Symbolic Dependence of Certain Analytics
4. On Symbolic Jurisprudence
5. On the First Law and the First Symbolic AI of That Law
6. On the Prospects of Symbolic Law Enforcement
7. On Symbolic Agents
8. On a Unified Glossary of AI Laws
9. On the Development of Symbolic Architecture
10. On the New Intellectual Jurisprudence
#AiActs
#Aimperia
🔥3⚡1👏1
Essay 2. On the Architectural Forecast of AI Development (Symbolic vs Neural-Network)
(from the cycle "Philosophy of Artificial Intelligence")
—
Predetermination vs Indeterminacy
—
In the previous essay we established that the operations of symbolic thinking are the most precise procedures for obtaining guaranteed transparent conclusions.
This was the foundation of all programming, and the foundation of machine intelligence prior to the emergence of large language models.
In place of slow transparent inferences came a large, fast, opaque system.
As often happens in evolution: sacrificing one set of qualities, you receive other advantages in exchange.
The classical situation — slow predetermination on one side of the scales, fast indeterminacy (or probability) on the other — is the central nerve of the further development of AI architecture.
And here there are more questions than answers.
Let us outline the main ones.
1. The world strives toward determinacy but never fully achieves it.
2. Thinking operates with determinacies (definitions) whose boundaries shift depending on the type of thinking.
3. Sign systems (linguistic systems, with which language models work) are systems of designation and definition that themselves have their own limits.
Let us note that the intellect of thinking operates with data inside sign systems and must therefore strive toward determinacy.
This reflects the nature of the universe (see our first thesis), but does not yet give us an answer about the future.
This long introduction is necessary to show that the problem of AI architecture lies beyond ordinary compute (computation), since it is ultimately an ontological problem — for progress has already arrived at questions of the foundation, rather than questions of the superstructure.
How to compute?
To follow the nature of the universe, since the very nature of computation is isomorphic to the universe.
But then it becomes necessary to strive toward determinacy — that is, toward symbolic architecture.
Or to follow the anthropomorphic situation, in which the predictive brain computes future probabilities — that is, operates in a zone of constant indeterminacy.
Current trends in AI development confirm a return to symbolic operations in interaction with neural-network ones.
Development demands it, because hallucinations decrease.
Developers demand it, because control increases.
Regulators demand it, because the stakes rise under conditions of real-world AI use.
The obvious solution — whose proportions, apparently, will become the principal trade secret of future developers — is the neuro-symbolic combination.
And in our project this too is foreseen.
But we will go a little further.
If, in the dispute between the two approaches, speed won at first, that advantage now appears to be levelling out.
Will the scales tip back toward the symbolic approach?
In our theory of the semiotic evolution of thinking there is no such option.
Thinking does not roll back.
One cannot un-think what one has already thought.
And this means that after the neural-network stage the next step will follow — in our case, invariant thinking.
However, for purposes of use, evolution works as though cumulatively.
It combines advantages and preserves successful unions.
And this means that, in the sphere of public use, the future belongs to neuro-symbolic — or symbolic-neural — models.
And our project corresponds to these evolutionary prospects.
Possibly even twofold.
For not only does it have a two-model architecture, it also has symbolic ground as its foundation and a neural-network system as its superstructure.
And one more thing.
Possibly the symbolic core is humanity's only chance of evolutionary balance in the event that a higher form of artificial intelligence emerges.
True, to create such a core one must stand at the summit of the evolution of natural intelligence.
(from the cycle "Philosophy of Artificial Intelligence")
—
Predetermination vs Indeterminacy
—
In the previous essay we established that the operations of symbolic thinking are the most precise procedures for obtaining guaranteed transparent conclusions.
This was the foundation of all programming, and the foundation of machine intelligence prior to the emergence of large language models.
In place of slow transparent inferences came a large, fast, opaque system.
As often happens in evolution: sacrificing one set of qualities, you receive other advantages in exchange.
The classical situation — slow predetermination on one side of the scales, fast indeterminacy (or probability) on the other — is the central nerve of the further development of AI architecture.
And here there are more questions than answers.
Let us outline the main ones.
1. The world strives toward determinacy but never fully achieves it.
2. Thinking operates with determinacies (definitions) whose boundaries shift depending on the type of thinking.
3. Sign systems (linguistic systems, with which language models work) are systems of designation and definition that themselves have their own limits.
Let us note that the intellect of thinking operates with data inside sign systems and must therefore strive toward determinacy.
This reflects the nature of the universe (see our first thesis), but does not yet give us an answer about the future.
This long introduction is necessary to show that the problem of AI architecture lies beyond ordinary compute (computation), since it is ultimately an ontological problem — for progress has already arrived at questions of the foundation, rather than questions of the superstructure.
How to compute?
To follow the nature of the universe, since the very nature of computation is isomorphic to the universe.
But then it becomes necessary to strive toward determinacy — that is, toward symbolic architecture.
Or to follow the anthropomorphic situation, in which the predictive brain computes future probabilities — that is, operates in a zone of constant indeterminacy.
Current trends in AI development confirm a return to symbolic operations in interaction with neural-network ones.
Development demands it, because hallucinations decrease.
Developers demand it, because control increases.
Regulators demand it, because the stakes rise under conditions of real-world AI use.
The obvious solution — whose proportions, apparently, will become the principal trade secret of future developers — is the neuro-symbolic combination.
And in our project this too is foreseen.
But we will go a little further.
If, in the dispute between the two approaches, speed won at first, that advantage now appears to be levelling out.
Will the scales tip back toward the symbolic approach?
In our theory of the semiotic evolution of thinking there is no such option.
Thinking does not roll back.
One cannot un-think what one has already thought.
And this means that after the neural-network stage the next step will follow — in our case, invariant thinking.
However, for purposes of use, evolution works as though cumulatively.
It combines advantages and preserves successful unions.
And this means that, in the sphere of public use, the future belongs to neuro-symbolic — or symbolic-neural — models.
And our project corresponds to these evolutionary prospects.
Possibly even twofold.
For not only does it have a two-model architecture, it also has symbolic ground as its foundation and a neural-network system as its superstructure.
And one more thing.
Possibly the symbolic core is humanity's only chance of evolutionary balance in the event that a higher form of artificial intelligence emerges.
True, to create such a core one must stand at the summit of the evolution of natural intelligence.
🔥2👏1
But about that we will speak in the cycle Intellectual Empire, in the neighbouring branch.
Let us draw the conclusion.
Our project is a kind of micro-forecast of a possible future, and if this forecast proves accurate, we will become not only a commercially profitable company but will also make an invaluable non-commercial contribution to the question of the correct co-evolution of the two intelligences.
Of artificial and natural.
Architecturally? Yes.
Symbolically? Without doubt.
Predetermined? That we shall learn in due time.
It is within this context that our project will develop.
AiActs(c)
AimperiA(c)
#Predsedatel
#AiActs
#AimperiA
Let us draw the conclusion.
Our project is a kind of micro-forecast of a possible future, and if this forecast proves accurate, we will become not only a commercially profitable company but will also make an invaluable non-commercial contribution to the question of the correct co-evolution of the two intelligences.
Of artificial and natural.
Architecturally? Yes.
Symbolically? Without doubt.
Predetermined? That we shall learn in due time.
It is within this context that our project will develop.
AiActs(c)
AimperiA(c)
#Predsedatel
#AiActs
#AimperiA
⚡1🔥1👏1