1.2 Describe How Various Data Types Are Used in Gen AI and the Business Implications. GENERATIVE-AI-LEADER Practice Questions (Page 5)

Questions 21–24

1. 21foundation · easy

   A company wants to use its customer support chat logs to train a generative AI model that answers customer questions. The logs contain many off-topic conversations and outdated information about discontinued products. Which data quality characteristic is most lacking in this dataset for the intended purpose?

   • AConsistency
   • BCompleteness
   • CAccuracy
   • DRelevance
   Reveal Solution Discussion Select an answer first
2. 22foundation · easy

   A data scientist is preparing a dataset for a generative AI model that will generate product descriptions. They notice that some product records are missing the 'color' field, and other records list the same product color in different formats (e.g., 'Blue' vs 'blue'). Which two data quality characteristics are most directly compromised in this dataset?

   • ACompleteness and consistency
   • BUniqueness and integrity
   • CAccuracy and validity
   • DRelevance and timeliness
   Reveal Solution Discussion Select an answer first
4. 23application · medium

   A manufacturing company wants to use gen AI to analyze sensor data from machines to predict maintenance needs. The sensor data is stored in a legacy database that is only accessible during business hours and requires a specialized team to export. The company wants to make the data more accessible for model training. Which action should they take?

   • AMove the sensor data to a cloud-based data warehouse that is accessible 24/7 and can be integrated with gen AI tools.
   • BContinue using the legacy database but schedule exports during business hours to avoid disrupting operations.
   • CUse a small sample of the sensor data for training to minimize the need for frequent access.
   • DHire more specialists to manually export the data as needed, ensuring the team has access when required.
   Reveal Solution Discussion Select an answer first
5. 24application · medium

   A legal firm wants to use gen AI to draft contract clauses. They have a large corpus of historical contracts, but only a few are annotated with the clause types (e.g., indemnity, termination). The firm wants to improve the model's ability to generate accurate clauses. Which approach should they take?

   • AUse the annotated contracts for supervised fine-tuning and the unannotated contracts for self-supervised learning to understand legal language patterns.
   • BDiscard the unannotated contracts to avoid introducing noise into the model.
   • CUse only the annotated contracts for training, as unannotated contracts lack the necessary labels for supervised learning.
   • DManually annotate all contracts to ensure the model has enough labeled examples for accurate clause generation.
   Reveal Solution Discussion Select an answer first