Governing AI design
AIGP · Lesson 10 of 15 · 1:24:10
YouTube loads only after you press play. This chapter opens at its timestamp in the full course. Watch on YouTube ↗
In this lesson
Prepare for the IAPP AIGP exam, on the AIGP Body of Knowledge version 2.1. In this lesson: how to govern the design of an AI system: the business context and a specific use case, whether AI is the right tool, impact assessments, requirements, architecture and model selection, levels of human oversight, thresholds, stakeholder engagement, operational controls, risk methods (probability and severity matrix, mitigation hierarchy, stakeholder mapping, benchmarks, pilots), and documentation. Includes three original practice questions, with Elmstead Group, a fictional retail and finance group, as the example.
Preview the topics
- Elmstead’s marketing team wants an AI system to choose which customers receive a new loyalty offer. Before the data science team starts, what should governance do first?
- Elmstead’s fraud model blocks card payments automatically, with no review before each block, but a specialist monitors its decisions in real time and can override them. Which level of human oversight is this?
- A model could reveal customers’ health data in its outputs. Elmstead can remove health data from training, add an output filter, or train staff to review outputs. Following the mitigation hierarchy, which should come first?