When conducting an AI system audit, what is the MOST important reason to review the model's feature engineering documentation?
Select an answer to reveal the explanation.
Short Explanation
Here's the deal — you review feature engineering documentation because it shows whether transformations may introduce bias or violate data-use restrictions. That's where proxy variables and protected-attribute handling can quietly create audit problems.
Full Explanation
The correct reason is that reviewing feature engineering documentation lets auditors assess whether feature transformations, such as aggregating protected attributes or using proxy variables, may introduce bias, and whether data was used consistently with collection consent and legal restrictions. The approved programming language option is not an audit objective. The minimum parameter count option is not an audit objective. The developer productivity option is a project management activity.