Skip to content
Get in touch

Reviewed in POC · content version 3

Review Capability Provenance

Your draft is bound to this exact source hash. If the term changes, the Review Desk will use a new storage key instead of silently attaching old decisions.

Definition

Capability provenance is the record of where an AI system's abilities came from — the models, data, human input, tools and processes that materially shaped what it can do.
Decision for Definition

Plain-Language Explanation

Capability provenance explains where a system's abilities came from. It connects behavior to material contributors such as training, fine-tuning, retrieval, tools, human feedback, integration, and operational practice.
Decision for Plain-Language Explanation

Why It Matters

Teams cannot evaluate control, ownership, risk, or reproducibility if they cannot trace the important contributors to capability.
Decision for Why It Matters

Example

A coding assistant's performance record identifies its base model, fine-tuning data, retrieval index, execution tools, reviewer feedback, and deployment configuration.
Decision for Example

Common Misconception

Capability provenance is not a claim that every output can be traced to one training example. It records material contributors to capability at an appropriate level.
Decision for Common Misconception

Remember It This Way

What made this system able to do that?
Decision for Remember It This Way

Related Terms

* [[evidence-selection|Evidence Selection]] * [[verification|Verification]] * [[layered-sovereignty|Layered Sovereignty]] * [[human-ownership|Human Ownership]]
Decision for Related Terms

Sources

No current wording.
Decision for Sources

Learn More

* Use with [[verification|Verification]] and [[evidence-selection|Evidence Selection]].
Decision for Learn More
Return to term

Not saved yet.