Key Metrics Explained

• AI Fairness Score: Measures overall fairness of AI models across protected attributes, ensuring equitable outcomes.
• Model Explainability: Percentage of model decisions that can be clearly explained to stakeholders.
• Guardrails Adherence: Measures adherence to data trust guardrails (privacy, ethics, robustness, transparency, accountability).
• GenAI Safety Index: Composite score reflecting the safety and reliability of generative AI models.
• Fairness Across Protected Attributes: Breakdown of fairness scores for demographic groups (e.g., gender, age).
• Model Explainability Across Features: Visualizes feature importance across AI models for transparency.
• Guardrails Adherence Radar: Multi-dimensional view of adherence to data trust principles.
• GenAI Safety Metrics: Tracks toxicity, bias, hallucination, privacy leakage, and factual accuracy for generative AI.
• AI Model Performance Over Time: Monitors accuracy, F1 score, and AUC-ROC to ensure consistent quality.

Built with Data Trust principles of agility and collaboration. See the Data Trust Manifesto for details on engineering rigor.