What metrics are important for AI governance?

• A. Metrics that monitor bias, fairness, transparency, and explainability.
• B. Only operational uptime.
• C. Only cost per model.
• D. Only time to deploy.

Answer: (A)

AI governance hinges on measuring how a model affects people and whether its decisions can be trusted. Metrics should capture bias and fairness, ensuring decisions are not unfair or discriminatory, and assess transparency so stakeholders understand how the model works. Explainability metrics matter because they reveal the reasoning behind outputs, enabling accountability and auditability. These governance-focused metrics address potential harm, regulatory compliance, and public trust, which operational metrics alone cannot. While uptime, cost per model, and time to deploy matter for operations, they do not address governance concerns about equity, understandability, and accountability.