Assess severity: What should guide the acceptance decision?
Low likelihood does not make severe harm acceptable; assess impact, mitigation, and documented residual-risk tolerance.
The question
In an EU public-benefit procurement, a vendor reports a very low error rate for an AI tool that recommends essential-service eligibility. A mistaken recommendation could seriously deprive a person of support. What should guide the acceptance decision?
Preparing for AIGP? Take the free 5-min readiness quiz →
- Require a lower supplier price before approving deployment.Price may affect procurement value, but it does not address the severity or treatment of eligibility errors.
- Provide applicants with a general system notice.Notice may support transparency, but it does not by itself prevent serious errors or provide effective correction and remedy.
- Approve it once average accuracy exceeds the target.Average accuracy can hide consequential subgroup or edge-case failures and does not establish acceptable residual risk.
- Assess severity, mitigation, and residual-risk tolerance. ✓A low error rate does not settle acceptance. The agency should assess the severity of possible deprivation, available mitigations, and whether documented residual risk is acceptable under applicable law and policy.
The trap
For rare but consequential failures, ask whether safeguards and acceptance criteria address the consequence, not merely the frequency. How to remember it
Low likelihood does not make severe harm acceptable; assess impact, mitigation, and documented residual-risk tolerance.
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 Development questions
- Rank risks by weighing likelihood and severity: Which distinction should govern prioritization? →
- Define acceptance criteria before reviewing results: What should it do first? →
- Claimant outcomes and distribution of delays: Which outcome should governance examine first? →
- All 426 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.