Examples
Worked examples
- Is an instance
A use card for a deployed clinical risk model documenting that it is to be used to triage referrals, not to make admission decisions.
- Is an instance
A use card stating an automatic-translation model is for first-draft assistance, not certified translation.
Counter-examples
Looks similar, but isn't
- Not an instance
A model card describing the underlying model.
- Not an instance
A marketing brochure.
Editorial commentary
A use card is a documentation artefact recording an intended deployment context for a model or system: the user population, the decisions the model is meant to inform, the human-supervision regime in place, and explicit out-of-scope uses the deployer will not support. Unlike most of the model-documentation family, a use card is typically written by the deploying organisation, not the model’s developer — it documents a specific application, not the model itself.
Where it sits in the documentation hierarchy
AI documentation runs from general method to specific deployment. An algorithm card describes a method class independent of any trained instance. A model card (Mitchell et al., 2019) describes one specific trained model — its training data, performance across subgroups, and intended/out-of-scope uses in general terms — written once by the model’s developer at release. A use card narrows that further to one particular deployment of that model inside one organisation: the same base model used for essay screening and for grant-abstract triage would reasonably have two different use cards, each written by the team deploying it in that context, even though both point back to the same underlying model card. A system card sits above all three, describing the full sociotechnical system a use card’s deployment is embedded in.
Why it matters for research governance
An IRB, procurement office, or research-integrity committee reviewing a proposed AI deployment needs the use card, not the model card alone — the model card tells them what the model can generally do; the use card tells them what a specific team intends to do with it and where human oversight sits.
References
- Mitchell, M. et al. (2019). “Model Cards for Model Reporting.” Proceedings of FAT* ’19.
- See also: Model card, System card, AI evaluation card.
Also known as
intended-use card · deployment card
Machine-readable encodings
Use in your systems
<role vocab="credit"
vocab-identifier="https://casrai.org/dictionary/"
vocab-term="Use card"
vocab-term-identifier="https://casrai.org/dictionary/term/use-card" />{
"@context": "https://schema.org",
"@type": "DefinedTerm",
"@id": "https://casrai.org/dictionary/term/use-card",
"name": "Use card",
"identifier": "https://casrai.org/dictionary/term/use-card",
"description": "A documentation artefact recording an intended deployment context for a model or system, including the user population, the decisions the model informs, the supervision regime, and out-of-scope uses.",
"inDefinedTermSet": "https://casrai.org/dictionary/domain/ai-ml-research-outputs#set",
"url": "https://casrai.org/dictionary/term/use-card",
"alternateName": [
"intended-use card",
"deployment card"
],
"license": "https://creativecommons.org/licenses/by/4.0/",
"publisher": {
"@id": "https://casrai.org/#organization"
},
"author": {
"@id": "https://casrai.org/#editorial-team"
},
"datePublished": "2026-05-21T02:22:50",
"dateModified": "2026-09-05T14:24:33",
"inLanguage": "en-GB",
"isAccessibleForFree": true
}
