A model card is a short document that an AI developer publishes with a model. It explains what the model is meant to do, how it was trained and tested, its known limits, and the uses it is not suited for. Think of the leaflet that comes with a medicine: it tells you what the product is for and when you should not use it. Some developers also publish a longer system card that covers a whole AI system or product rather than a single model.
Why it matters
If you are choosing an AI tool for your organization, the model card is a good first place to look. Read the parts on limits and unsuitable uses, and compare them with what you plan to do, such as summarizing case notes or answering questions from the public. Keep in mind that the card is the developer’s own account, not an independent review, so treat it as a starting point and not the final word. A fair question to ask a vendor is: “Which model does your product use, and can you share its model card?” If they cannot answer, that is useful to know. For how models are tested, see model evaluation.
Published Updated