1.4 AI/Machine Learning 400-007 Practice Questions (Page 4)

Questions 16–20

1. 16foundation · easy

   Which statement best describes the primary purpose of identifying business needs before adopting AI/ML solutions?

   • ATo select the most advanced AI/ML algorithms available on the market
   • BTo comply with regulatory requirements for AI/ML adoption
   • CTo ensure the AI/ML solution aligns with organizational goals and delivers measurable value
   • DTo reduce the need for ongoing IT infrastructure investments
   Reveal Solution Discussion Select an answer first
2. 17expert · hard

   A global bank is building an AI model to detect money laundering. The model uses transaction data from multiple countries, and the bank must ensure that the data used for training is consistent and has not been tampered with. The bank also needs to prove to regulators that the data has not been altered. Which combination of mechanisms should the bank implement?

   • AStore data in a secure data warehouse with role-based access controls.
   • BRely on the cloud provider's data encryption and backup services.
   • CUse cryptographic hashing of data files and store the hashes in a separate secure location.
   • DUse a blockchain-based data provenance system to record every data access and modification.
   Reveal Solution Discussion Select an answer first
4. 18foundation · easy

   Which element is a key component of AI/ML governance?

   • AChoosing the most accurate algorithm for the dataset
   • BOptimizing network bandwidth for data transfer
   • CLifecycle management policies for models from development to retirement
   • DSelecting the appropriate GPU type for training
   Reveal Solution Discussion Select an answer first
5. 19expert · hard

   A telecommunications company has deployed an AI model for network anomaly detection. The model runs on a distributed cluster and processes streaming data from network sensors. The operations team notices that the model's performance degrades during peak traffic hours, and they suspect that the data distribution shifts. The team needs to implement an assurance mechanism that can detect this shift in near real-time without adding significant latency to the inference pipeline. Which approach should they choose?

   • AAdd more inference nodes to handle the peak traffic, assuming that performance degradation is due to capacity.
   • BIncrease the model's complexity to make it more robust to distribution shifts.
   • CRetrain the model every hour using the latest streaming data to ensure it adapts to changes.
   • DImplement a separate monitoring service that samples a subset of incoming data and compares the distribution to the training data distribution using a statistical test.
   Reveal Solution Discussion Select an answer first
6. 20expert · hard

   A hospital is deploying an AI system to predict patient readmission risk. The system uses electronic health records (EHR) and must comply with strict privacy regulations. The hospital also needs to ensure that the model's decisions are explainable to clinicians and auditors. The data science team proposes using a deep learning model with high accuracy but low interpretability. What should the design team do?

   • AUse the deep learning model but restrict access to the model's outputs to only a few experts.
   • BApprove the deep learning model and rely on post-hoc explanation tools to provide interpretability.
   • CReject the deep learning model and require a simpler, interpretable model such as logistic regression, even if accuracy is slightly lower.
   • DDeploy the deep learning model and document that the model is a black box, accepting the risk.
   Reveal Solution Discussion Select an answer first