4.3 Describe the Importance of Responsible AI in Business. GENERATIVE-AI-LEADER Practice Questions (Page 3)

Questions 11–15

1. 11expert · hard

   A pharmaceutical company is using a generative AI system to assist in drug discovery. The system suggests molecular structures, but researchers are unsure why certain structures are suggested. The company wants to publish results in a scientific journal, which requires methodological transparency. What is the best way to meet this requirement?

   • APublish only the successful results without detailing the AI process
   • BKeep the model’s inner workings confidential to protect intellectual property
   • CUse a simpler model that is easier to explain, even if it is less effective
   • DDocument the model’s architecture, training data, and provide explanations for its suggestions
   Reveal Solution Discussion Select an answer first
2. 12foundation · medium

   How does explainability help stakeholders trust AI decisions?

   • ABy ensuring the AI always makes the same decision for the same input.
   • BBy providing reasons for the AI's decisions that stakeholders can evaluate.
   • CBy hiding the decision-making process to avoid confusion.
   • DBy making the AI system's code available for review.
   Reveal Solution Discussion Select an answer first
4. 13application · medium

   A logistics company trains a generative AI model to draft customer service responses. The training data includes many outdated product names and incorrect addresses. The model now generates responses with wrong information. What is the most likely business impact?

   • AHigher model accuracy on new products
   • BReduced need for human oversight
   • CDecreased customer trust and increased support costs
   • DImproved response speed due to fewer data points
   Reveal Solution Discussion Select an answer first
5. 14expert · hard

   A data analytics company is preparing a dataset for training a generative AI model. The dataset contains customer purchase histories. The company wants to share the dataset with a third-party vendor for model training while minimizing privacy risks. They plan to replace customer names with unique IDs but keep purchase details. A privacy consultant says this is insufficient. What is the best next step?

   • AObtain consent from all customers before sharing the data
   • BRemove purchase details and keep only aggregate statistics
   • CAdd random noise to the purchase amounts to make them less precise
   • DApply k-anonymity to ensure each record is indistinguishable from at least k-1 other records
   Reveal Solution Discussion Select an answer first
6. 15expert · hard

   A research consortium is using a public generative AI API to analyze genomic data for disease research. The data is highly sensitive and can be re-identified even after removing direct identifiers. The consortium wants to use the API for its advanced capabilities but must protect patient privacy. Which approach best balances utility and privacy?

   • ASend the raw genomic data to the API and rely on the API provider’s privacy policy
   • BUse differential privacy techniques to add noise to the data before sending it to the API
   • CRemove all direct identifiers and send the data, assuming it is anonymized
   • DDo not use the API and instead use a less capable on-premises model
   Reveal Solution Discussion Select an answer first