To evaluate the potential harms the system could pose: What is the primary purpose of this activity?
The primary purpose of an impact assessment is to identify and evaluate the potential harms the AI system could pose to affected individuals and groups.
The question
A deployer performs an impact assessment on a selected AI system before go-live. What is the primary purpose of this activity?
Preparing for AIGP? Take the free 5-min readiness quiz →
- To evaluate the potential harms the system could pose to affected people ✓An impact assessment systematically identifies and evaluates the potential harms of the system to affected individuals and groups.
- To negotiate favorable pricing and support tiers with the model vendorA real procurement objective, but commercial negotiation is unrelated to assessing the system's potential harms.
- To schedule the recurring retraining and maintenance windows after launchA valid operational plan, yet scheduling upkeep is separate from evaluating the harms the system may cause.
- To confirm the network latency budget for the chosen hosting locationA legitimate engineering check, but latency planning does not assess the system's impact on people.
The trap
Confusing an impact assessment with commercial, operational or engineering tasks that occur around deployment. How to remember it
The primary purpose of an impact assessment is to identify and evaluate the potential harms the AI system could pose to affected individuals and groups.
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
- An edge-deployed model fine-tuned on the company's domain: Which combined approach best fits all constraints? →
- Identifying and evaluating key terms and risks: Which assessment activity is this? →
- It carries greater direct accountability and higher: Which of the following is a risk that is distinctly →
- 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.