Add controls proportionate to autonomous rejection: With the same model and data, what governance change is
Automatic rejection increases autonomy and impact, requiring stronger oversight despite unchanged technology.
The question
A university research office first uses an AI tool to flag incomplete grant applications for staff review. It later authorizes the tool to reject applications automatically and notify applicants. With the same model and data, what governance change is most important?
Preparing for AIGP? Take the free 5-min readiness quiz →
- Add controls proportionate to autonomous rejection, including accountable review and impact monitoring. ✓Automatic rejection increases autonomy and applicant impact, requiring stronger oversight, accountability, and monitoring despite the unchanged model and data.
- Monitor rejection accuracy and notification speed after launch.Performance monitoring is relevant, but accuracy and speed alone do not provide oversight for consequential automated rejection.
- Rely on the unchanged model version.The model version does not determine the governance exposure by itself; the system now exercises more consequential authority.
- Retain the original pilot controls and document the change.Documentation is useful, but retaining the original controls is insufficient when the system can now reject applications without staff action.
The trap
Compare the system’s authority and consequences, not merely its model version or data. How to remember it
Automatic rejection increases autonomy and impact, requiring stronger oversight despite unchanged technology.
How many of these would you get right?
One of 1581 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 the Foundations of AI Governance questions
- Use the probabilities as decision support with calibrated: Which governance response is most appropriate? →
- Record individual and group impacts separately: Which distinction should reviewers record? →
- Treat reliability as unestablished because performance: Which conclusion is most defensible? →
- All 337 Understanding the Foundations of AI Governance questions →
Part of the Certsqill AIGP question bank · Understanding the Foundations of AI Governance ·
Every answer, right and wrong, comes with its own explanation.