Question 19
Domain 2 — AI Operations, Lifecycle, and Control EnvironmentWhich of the following is MOST important to review in order to gain assurance that an AI model is performing without biases?
Correct answer: C
Explanation
AI training data is the key input that shapes model behavior, so reviewing it helps identify skewed, incomplete, or unrepresentative examples that can produce biased outputs. Bias often originates in the data used to train the model, making the training set the most important area to inspect for fairness and assurance.
Why each option is right or wrong
A. AI model temperature
B. AI development environment
C. AI training data
AI model bias is most often introduced at the point of model development, so the primary control to inspect is the training dataset used to fit the model. Under NIST AI RMF 1.0, data quality and representativeness are core governance considerations, and ISO/IEC 23894:2023 likewise treats biased or incomplete data as a principal source of AI risk; reviewing the training data helps detect underrepresentation, label skew, and historical prejudice before those patterns are learned by the model.
D. AI model adaptability