“Maria describes the open-source library Giskard as a tool for evaluating AI models, specifically for detecting bias, running tests, and generating trust metrics.”