A company deploying ML models on AWS wants to provide transparency into model decision-making and offer explanations for predictions. Which AWS feature supports this need?

Amazon SageMaker Model Cards provide a centralized, standardized way to document ML models throughout their lifecycle. This includes capturing details about model purpose, training data, performance metrics, and most importantly, explanations for model predictions. This transparency helps stakeholders understand how a model arrives at its decisions, which is crucial for ethical AI and regulatory compliance. Amazon Rekognition is a service for image and video analysis, not for general model explanation. Amazon Comprehend is a natural language processing (NLP) service, and while it can explain its own NLP predictions, it doesn't offer general model explanation for arbitrary ML models. Amazon Lex is a service for building conversational interfaces (chatbots), not for model transparency.