Require policy-source verification and escalation when: Which mitigation addresses the remaining risk most
Training awareness has not changed behavior; mandatory source verification and escalation directly address confident unsupported answers.
The question
An education administration chatbot answers enrollment-policy questions for staff. Operators complete training explaining that generated answers may be confident but unsupported. A pilot still shows staff accepting answers without checking cited policy text, especially during peak enrollment. Which mitigation addresses the remaining risk most directly?
Preparing for AIGP? Take the free 5-min readiness quiz →
- Restrict the chatbot to frequently asked questions documented by administrators.A narrower scope may reduce exposure but cannot ensure staff detect unsupported answers or escalate conflicts within the remaining questions.
- Require policy-source verification and escalation when evidence is missing or conflicting. ✓This converts awareness into a concrete control addressing unsupported outputs, citation checking, uncertainty, and human escalation during consequential work.
- Display a larger warning that chatbot answers are informational only.A warning may communicate limitations but does not ensure staff verify policy evidence or escalate uncertain answers during peak demand.
- Increase the frequency of operator training sessions.More training may reinforce awareness, but it does not require verification behavior when staff face confident unsupported answers under pressure.
The trap
If users know the limitation but still accept outputs, add a required behavior—verification, abstention, or escalation. How to remember it
Training awareness has not changed behavior; mandatory source verification and escalation directly address confident unsupported answers.
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 How to Govern AI Deployment and Use questions
- Preserve versioned prompts: Which evidence should the service preserve to support comparison and corrective →
- Silent subgroup errors may remain undetected: Which specific risk remains insufficiently addressed by that →
- Reviewers may rubber-stamp outputs under deadline pressure: Which remaining risk is most specific? →
- All 424 Understanding How to Govern AI Deployment and Use questions →
Part of the Certsqill AIGP question bank · Understanding How to Govern AI Deployment and Use ·
Every answer, right and wrong, comes with its own explanation.
