Architecture /
A Common Language for Governed AI

Private Overview

EN · 中文

LISSAI's Canonical Glossary establishes a consistent vocabulary for meaning, authority, constraints, and governance across its architecture.

Canonical Glossary

LISSAI's architecture is supported by a deliberately developed vocabulary that defines its core concepts, governing principles, and functional relationships. Many of these definitions have been developed specifically for LISSAI, while others clarify established terminology within the context of governed AI systems.

This shared language helps preserve precision and consistency across architectural development, verification, and communication. By establishing clear distinctions between meaning, authority, inference, and execution, the Canonical Glossary provides a common reference for understanding LISSAI without conflating its functions with those of conventional AI systems.

Sample LISSAI Canonical Glossary entries (abbreviated)

  • Human-authored definitions of context, intent, and semantic structure that establish how information is understood and governed within LISSAI. Meaning is explicitly defined before AI inference, rather than independently determined or redefined by the model.

  • A framework that evaluates whether real-world operational requirements are sufficiently defined to become enforceable authority. It assesses the completeness of context, scope, constraints, evidence, and risk before authority is formally delegated.

  • A structured, human-authored agreement that defines what an AI system is permitted to do. It establishes the scope of authority, applicable constraints, governance requirements, and conditions for escalation before consequential actions may be authorized.

  • The governance mechanism through which LISSAI establishes enforceable relationships between meaning, authority, and execution constraints. It provides the foundation for deterministic authorization and accountability in consequential AI operations.

  • A process designed to transform human-defined meaning and authority into structured, machine-readable representations without losing intended meaning or governance constraints. It supports semantic consistency as information moves through the system.

  • A deterministic verification process that evaluates whether a proposed AI action is authorized and admissible before execution. PRVI verifies compliance with established governance conditions, not whether the action or its underlying information is correct in the real world.

  • The cryptographic security framework designed to protect semantic integrity, authorization, and provenance. It supports verification that governed information and authority remain authentic and have not been altered without authorization.

  • The gradual degradation or divergence of intended meaning as information is reinterpreted, reused, or influenced by changing contexts over time. Semantic Corrosion can occur even when an AI system continues to function normally, creating risks to consistency, authority, and governance integrity.

« «