AIGP - Understanding how to govern AI development (27% of the exam) - Section 3.1

Govern the design and build of an AI system by defining the business context, performing impact assessments, applying ethical considerations, and documenting the process for compliance and risk management.

Govern AI system design and build by defining business context, performing impact assessments, applying ethics by design, and documenting model selection decisions for compliance. Recognise how thorough use case assessment at the outset reduces downstream risk and rework.

AI impact assessmentethics by designmodel selectionuse case assessment

Practice question for this objective

Free sampleUnderstanding how to govern AI developmenthard

A retailer is starting to build an agentic AI system that will plan and execute multi-step refund workflows, calling internal tools to issue payments and update records with limited human review. The product owner is eager to begin model selection and prompt engineering. Before any of that, the governance lead insists the team first nail down the business context for the use case. Which task most directly establishes that business context at the design stage?

AArticulate the business problem the system must solve, the value it is expected to deliver, the decisions and actions it will take, and the constraints and affected parties that bound it. Correct
BBenchmark several candidate foundation models on a held-out refund-reasoning dataset so the team can pick the most accurate one before scoping the workflow.
CDraft the post-deployment monitoring metrics and alert thresholds the operations team will watch once the refund agent is running live in production.
DWrite the user-facing transparency notice telling customers that an automated system, rather than a human agent, is handling their refund request end to end.

Framing an AI use case begins with defining the business problem, expected value, the actions the system will take, and the constraints and affected parties before model selection. Business context for an AI use case is established by stating the problem to solve, the value sought, the decisions and actions the system will take, and the constraints and stakeholders that bound it. For an agentic system that acts autonomously, this framing is what later model selection, risk assessment and controls all build on, so it must precede benchmarking, monitoring design or transparency drafting.

Why A is correct: Defining the problem, expected value, the actions the agentic system will take autonomously, and the constraints and affected parties is exactly what framing the business context means; it sets the purpose and boundaries that every later design and risk decision depends on.

Why B is wrong: Comparative benchmarking is a real and useful activity, but it is model selection, which presupposes the business context is already defined; choosing a model before the problem, value and constraints are framed risks optimising for the wrong objective.

Why C is wrong: Monitoring metrics matter, but they belong to operations once the system is live and cannot be set sensibly until the use case and its objectives are defined; this skips the framing the governance lead is asking for.

Why D is wrong: A transparency notice is a legitimate downstream obligation, but it is a disclosure control that follows from design decisions; it does not establish the business problem, value, or constraints that frame the use case.

See more AIGP practice questions, answers explained.

Exam traps in Understanding how to govern AI development

Answers that look right on this material and are not. Each one is a distractor from a different question in the AIGP bank for this domain.

Compare candidate model architectures on accuracy and interpretability and shortlist the algorithm whose performance profile best fits the loan-prioritisation task at hand.
Why it is wrong: Tempting because it sounds like early design work, but comparing architectures is model selection, which the business context is meant to precede and inform, not replace.
Schedule a single ethical sign-off review just before launch so an independent panel can confirm the finished system reflects the values agreed in the workshop.
Why it is wrong: A pre-launch review has value, but relying on a single late sign-off is precisely the gap ethics by design exists to close; by launch the design is largely fixed, so the ethical requirements were not actually built in along the way.
It defers all ethical judgements to a specialist committee that reviews the finished AI system shortly before its scheduled release date and sign-off.
Why it is wrong: This is tempting because committee review feels rigorous, but reviewing a finished system is precisely the late checkpoint that ethics by design is meant to supplement, not the proactive embedding the concept describes.

Examworthy is not affiliated with or endorsed by IAPP. Original, blueprint-aligned practice material only.