What are Model Cards?

• A. Structured documents providing technical details, intended use, and limitations of an AI model
• B. A dataset labeling technique
• C. A data privacy policy
• D. A deployment checklist

Answer: (A)

Model cards are structured documents that summarize a machine learning model’s technical details, intended uses, and limitations. They explain what the model is designed to do, who it’s for, and what kinds of inputs and outputs it handles. They also document performance metrics across relevant datasets, fairness or bias considerations, and known limitations or potential risks. In addition, they outline the data sources used for training and evaluation, deployment considerations (like monitoring and update plans), and cautions about environments where the model may not perform well. The goal is to provide transparency so developers, governance teams, and stakeholders can assess suitability, manage risk, and decide where and how the model should be deployed. This makes them distinct from labeling techniques, data privacy policies, or deployment checklists.