4.4 Clause
Understanding AI methods and AI system capabilities
The subclause explains the difference between a method and a capability. A method is how the system does something; a capability is what it can do. The same capability can be delivered by more than one method, and one system can hold several methods at once.
What this settles
This is the distinction that makes the matrix useful. Teams that collapse it into one list lose the ability to ask the interesting question: if we change the method, which capability claim is at risk?
Who it binds
The organisation, as a matter of method. The distinction is not a legal category, but it is the grammar the rest of the framework uses.
What to do in your practice
For one live system, write two columns: the methods actually used and the capabilities actually claimed. Then draw the links. A capability with no method behind it is a claim without a mechanism; a method with no capability is cost without a purpose.
Evidence an auditor or regulator asks for
| A two-column method and capability sheet for one live system. | Tests the distinction on real work. |
|---|---|
| A link from each claimed capability to the method that delivers it. | Shows the claim has a mechanism. |
| A note of capabilities that no method supports. | Surfaces over-claiming early. |
For each capability we claim, which method delivers it, and would the claim survive a change of method?
Common gaps this page catches
- The inventory lists models, not methods and capabilities, so the two are confused.
- A capability is claimed because a vendor says so, with no method recorded.
- A method is recorded without saying what capability it serves.
What to prepare before publication
Run the two-column exercise on the system with the most vendor claims. It usually finds both kinds of gap in an afternoon.
How to test it here
Take one live system rather than a template, find the field or sentence where Understanding AI methods and AI system capabilities is used, and put this page beside it. Record the answer, the artefact and the date. A glossary entry on its own is an intention; a dated record beside a live artefact is evidence.
What this page does not claim
The page does not reproduce the draft, does not know the descriptor wording that the public preview withholds, and does not certify the system or the organisation. It states the unit at the claim level "concepts clause" and stops where the public record stops.
Common questions
- What does Understanding AI methods and AI system capabilities mean in ISO/IEC DIS 42102?
- The subclause explains the difference between a method and a capability. A method is how the system does something; a capability is what it can do. The same capability can be delivered by more than one method, and one system can hold several methods at once.
- What evidence should we keep for 4.4?
- A two-column method and capability sheet for one live system. A link from each claimed capability to the method that delivers it.
- Is Understanding AI methods and AI system capabilities a requirement?
- No. The draft is a descriptive framework. It sets no management system requirement and no conformity test. It becomes binding only when a contract, a regulator or a management system adopts it.
Related units
Last updated: · Markdown version · Check your system