Bias CT-AI Practice Questions (Page 4)

Questions 16–20

1. 16foundation · easy

   Which of the following is a basic technique to mitigate bias during model development?

   • ADeploying the model without testing
   • BUsing fairness-aware algorithms that incorporate fairness constraints
   • CIncreasing the number of training epochs
   • DUsing a larger batch size
   Reveal Solution Discussion Select an answer first
2. 17foundation · easy

   Which type of bias is characterized by the systematic omission of certain groups from the training data?

   • AMeasurement bias
   • BExclusion bias
   • CSample bias
   • DPrejudice bias
   Reveal Solution Discussion Select an answer first
4. 18foundation · easy

   What type of bias occurs when the training data does not accurately represent the population the model will serve?

   • AMeasurement bias
   • BSample bias
   • CPrejudice bias
   • DExclusion bias
   Reveal Solution Discussion Select an answer first
5. 19application · medium

   A news organization is using an AI system to recommend articles to readers. The system was trained on click data from the past year. The data shows that readers from certain demographic groups click on different types of articles. The system now recommends articles that reinforce the existing preferences of each group. The organization's editor is concerned that this creates 'filter bubbles' and reduces the diversity of content readers see. Which impact of bias is most directly demonstrated?

   • AIncreased legal liability for the news organization
   • BIncreased computational cost due to the recommendations
   • CReduced user trust due to a lack of diverse content
   • DReduced model accuracy for all readers
   Reveal Solution Discussion Select an answer first
6. 20application · medium

   A government agency is using an AI system to flag fraudulent benefit claims. The system was trained on data from past investigations. The investigators, due to limited resources, only investigated claims that were flagged by an older, rule-based system. The new AI system now learns to mimic the older system's biases. Which source of bias is most directly demonstrated?

   • AMeasurement bias, because the fraud labels were not accurately recorded
   • BData bias, because the training labels were generated by a biased process
   • CHuman interaction bias, because the investigators intentionally introduced bias
   • DAlgorithm bias, because the new AI system's algorithm is flawed
   Reveal Solution Discussion Select an answer first