Enabling a qualified person to review: When applying best practices to designing and building an AI system,
Human oversight means a qualified person can review, override or halt the system's outputs before they affect an individual.
The question
When applying best practices to designing and building an AI system, which measure most directly provides human oversight of the system's decisions?
Preparing for AIGP? Take the free 5-min readiness quiz →
- Enabling a qualified person to review, override or halt the system's outputs before they take effect on an affected individual. ✓Correct because human oversight means a competent person can monitor, intervene in, override or stop the system's decisions, which is precisely this measure.
- Compressing the trained model so it runs faster on edge hardware and returns its predictions with noticeably lower end-to-end latency.Plausible because performance optimization is a real build activity, but reducing latency concerns efficiency, not giving humans control over the system's decisions.
- Encrypting the model's stored parameters at rest so that unauthorized parties cannot extract or tamper with the deployed system's weights.Plausible because parameter security is a genuine control, but encryption protects confidentiality and integrity rather than providing human oversight of outputs.
- Expanding the training dataset with additional labeled examples so the model generalizes better across the range of inputs it will encounter.Plausible because more representative data can improve quality, but enlarging the dataset addresses accuracy, not the ability of humans to review and override decisions.
The trap
Equating any good build practice with human oversight, when oversight specifically means a person can review, override, or stop the system's decisions. How to remember it
Human oversight means a qualified person can review, override or halt the system's outputs before they affect an individual.
How many of these would you get right?
One of 499 AIGP questions on Certsqill. Take a free five-minute check and see your score per domain — not one number, but which section to open tonight.
Test your AIGP readiness — freeMore Understanding How to Govern AI Development questions
- Log the incident in a central register with root cause: Which response best discharges the obligation to →
- Bias testing, which examines the model's outputs across: Which test type is specifically intended to detect →
- Instructions for use provided to deployers: Which item is a public/transparency disclosure aligned with that →
- All 135 Understanding How to Govern AI Development questions →
Part of the Certsqill AIGP question bank · Understanding How to Govern AI Development ·
Every answer, right and wrong, comes with its own explanation.
