A customer wants a scalable framework for governing AI systems across many teams. Which set of pillars should the framework be organised around?
Select an answer to reveal the explanation.
Short Explanation
You need a governance framework that scales, so organise it around fairness, explainability, robustness, transparency, and accountability. Apply those same pillars to every model in scope. Accuracy, local controls, or cost and latency don't give you consistent governance.
Full Explanation
The correct set of pillars is fairness, explainability, robustness, transparency, and accountability, applied consistently to every model in scope. These give a common structure that scales across teams because every model, regardless of use case, can be assessed against the same five questions.
The model-accuracy-alone option is incorrect because accuracy is a performance metric, not a governance one; a highly accurate model can still be biased, opaque, and unaccountable.
The option of using whatever controls each team already has in place, left unharmonised across the organisation, is incorrect because leaving controls unharmonised defeats the purpose of an organisation-wide framework and produces inconsistent risk coverage.
The cost-and-latency-targets option is incorrect because cost and latency are operational concerns that sit outside governance's core scope.
For the exam, know the five governance pillars by name and be ready to map a scenario onto the one it tests.