An AI audit of a document classification system reveals the model achieves 95% accuracy overall but only 71% accuracy on documents from a specific business unit. What audit follow-up action is MOST appropriate?
Select an answer to reveal the explanation.
Short Explanation
Here's the deal — you need to investigate whether the accuracy gap comes from distributional differences in that unit's documents and decide if use there should be restricted or the model retrained. The 24-point gap for a subpopulation is too big to ignore.
Full Explanation
Investigating distributional differences and deciding whether to restrict that unit's use or retrain is right because the 24-point subpopulation gap needs root-cause review, such as underrepresented training data, plus a risk-based remediation decision. Accepting high overall accuracy ignores that failure. Suspending the model and manually reviewing all records is disproportionate. Reporting it as a cybersecurity incident misclassifies the issue.