Preface: Recognition of the Problem
At first, we studied AI document-writing technology by focusing on the innovative potential of a method in which artificial intelligence automatically drafts briefs and legal documents.
However, in the course of research, we confirmed that because artificial intelligence is a technology with autonomy, errors, distortions, omissions, and logical leaps may occur, and that artificial intelligence itself bears no legal responsibility for those results.
Therefore, apart from its technical possibility, a method in which artificial intelligence directly drafts legal documents contains legal and institutional risks, and promoting it simply as innovation arises from a lack of structural understanding of AI autonomy.
Accordingly, we shifted the direction of research away from a technology that allows artificial intelligence to write documents on behalf of users and toward a system that verifies the structure between existing documents and judgment materials.
The result is the MAAPCS System, designed in 17 languages and operating globally as the Structure Alignment Protocol for AI-based Verification Systems.
Chapter 1: Fundamental Limits of Artificial Intelligence
1.1 The Two Faces of Autonomy
The greatest characteristic of artificial intelligence is autonomy. However, this autonomy also becomes the cause of the following structural errors.
- Error: misunderstanding of facts and generation of inaccurate information
- Distortion: amplification of bias in training data
- Omission: loss of important facts or context
- Logical Leap: inference lacking logical validity
1.2 The Accountability Gap
The more fundamental problem is the attribution of responsibility. Artificial intelligence cannot be a subject of legal responsibility, and therefore it does not bear responsibility even when any error occurs. This creates a critical problem in the legal field. Efficiency in judgment may increase, but when an error occurs, responsibility is left in a vacant state, and this structure is incompatible with the rule of law.
Chapter 2: Shift in Research Direction
2.1 A New Paradigm
We sought a new paradigm for objectively verifying the structure of documents and judgment materials.
2.2 Transition to a Verification System
The core principles of this shift in research direction are as follows.
- Judgment belongs exclusively to humans or institutions.
- The function of artificial intelligence is limited to verification and analysis.
- Responsibility is always fixed to humans or institutions.
Chapter 3: Birth of MAAPCS
3.1 System Overview and Philosophy
MAAPCS is a structural verification system designed to secure objectivity and reliability in legal judgments and administrative acts in the age of artificial intelligence. This system limits the autonomy of artificial intelligence to technical functions and is designed so that final judgment and responsibility belong only to humans or lawful institutions.
3.2 The World’s First AI Autonomy Control System
MAAPCS is the world’s first system that controls the autonomy of artificial intelligence through structures and rules.
3.3 Patent Application and Academic Verification
MAAPCS has patent applications filed for its three-stage evolutionary structure, and academic verification is being conducted through research papers.
Chapter 4: Core Structure of MAAPCS
4.1 Four Principles of Structural Control
- Principle of Fixed Judgment Structure: judgment is performed only by humans or institutions as the stage of final conclusion, and artificial intelligence cannot participate in it.
Principle of Fixed Judgment Structure: judgment is performed only by humans or institutions as the stage of final conclusion, and artificial intelligence cannot participate in it.
- Principle of Limiting the Scope of AI Functions: artificial intelligence performs only data processing and structural analysis, and functions such as drafting legal documents, selection, and conclusion drawing are excluded.
Principle of Limiting the Scope of AI Functions: artificial intelligence performs only data processing and structural analysis, and functions such as drafting legal documents, selection, and conclusion drawing are excluded.
- Principle of Fixed Objective Function and Procedure: verification standards and procedures are defined externally in advance and cannot be changed by artificial intelligence.
Principle of Fixed Objective Function and Procedure: verification standards and procedures are defined externally in advance and cannot be changed by artificial intelligence.
- Principle of Fixed Attribution of Responsibility: all outputs of artificial intelligence are reference materials without legal effect, and responsibility is attributed entirely to humans or institutions.
Principle of Fixed Attribution of Responsibility: all outputs of artificial intelligence are reference materials without legal effect, and responsibility is attributed entirely to humans or institutions.
4.2 Dual-Layer System
- Computation Layer: artificial intelligence performs data processing and structural analysis, creating reference materials.
- Judgment Layer: humans or institutions perform final judgment, in complete separation. Every judgment process is recorded for audit tracking.
MAAPCS does not perform a function of generating or selecting legal documents based on user input. It is a system that verifies only structural consistency in existing documents. Therefore, it does not constitute the drafting of documents concerning legal relationships in a case or the handling of any other legal affairs.
Chapter 5: Significance and Consistency of MAAPCS
5.1 Consistency with International Standards
- The UNESCO Recommendation on the Ethics of Artificial Intelligence takes human control, accountability, and transparency as core principles.
- The OECD Recommendation on Artificial Intelligence requires human-centeredness and responsible innovation.
- The Framework Act on Artificial Intelligence of the Republic of Korea requires risk control for high-risk AI.
- By structurally controlling the autonomy of artificial intelligence, MAAPCS conforms to ISO/IEC technical standards requiring implementability, verifiability, and reproducibility.
5.2 Consistency with Laws
Article 202 of the Civil Procedure Act of the Republic of Korea and Article 307 of the Criminal Procedure Act presuppose structural completeness at the pre-judgment stage. MAAPCS is a system that verifies the existence of that structural completeness and follows the same framework as the judgment structure required by law. Therefore, this verification does not replace legal judgment but constitutes a technical record verifying the foundational structure of judgment.
Conclusion
In the course of research on AI document-writing technology, we recognized the fundamental risks of AI autonomy and shifted the direction of research. The resulting MAAPCS is a verification system of a new paradigm that controls the autonomy of artificial intelligence and fixes judgment and responsibility to humans. MAAPCS is not a technology that performs legal affairs, but a technology that verifies the structural consistency of legal judgment, and it is an essential foundation for securing legal stability in the age of artificial intelligence.