Model Cards

Short structured documents, proposed by Mitchell et al. (2019), that accompany a machine learning model and report its intended uses, performance across relevant demographic subgroups, training data, evaluation metrics, known limitations, and ethical considerations. Model cards are intended to democratise AI transparency by giving non-expert audiences — including disabled users, regulators, and affected communities — enough information to judge whether a model is appropriate for a given context. In public-service AI deployments, they are increasingly proposed as a baseline accountability mechanism alongside open-source code and external audit.