Artificial Intelligence Architecture

Kind: model

Layer: Correctness Core

Record: architecture:artificial-intelligence-architecture

Severity: contextual

Scope: model-backed system, application, platform

Canonical: Ontology

A conceptual representation of a system that calls a model behind validated inputs, grounded context, schema-checked outputs and governance.

Listed in Architecture principles, after Model-Driven Architecture and before Machine Learning Architecture.

Requires

• Model Governance
• Data/Model Boundaries

Reinforces

• Model Evaluation
• Model Safety

Enables

• Model-Integrated Systems

Conflicts with

• Opaque Ungoverned Model Use

In tension with

• Determinism
• Explainability

Tensions

• Artificial Intelligence Architecture / Determinism
• Artificial Intelligence Architecture / Explainability

Violated by

• Opaque Ungoverned Model Use

Refactored by

• Evaluation Suite
• Introduce Port
• Guardrails

Severity

• contextual

Category

• Model Architecture

Reinforced by

• Model Inference

Linked from

• Correctness / Determinism / Verification
• Model Architecture
• Codebase / System Architecture Styles
• Model Architecture
• Severity levels
• The resolutions