Test Environments for AI-Based Systems CT-AI Practice Questions (Page 5)

Questions 21–25

1. 21foundation · easy

   Which of the following is a key component of an AI test environment?

   • ASource code editor
   • BEmail server
   • CWeb browser
   • DData pipelines
   Reveal Solution Discussion Select an answer first
2. 22expert · hard

   A large organization is testing an AI model that will be deployed in a customer-facing application. The test environment is shared with other teams, and the model's performance is affected by other teams' workloads. The team needs reliable performance metrics and wants to integrate testing into their CI/CD pipeline. What is the best approach?

   • AUse the production environment for performance testing to get real-world metrics.
   • BProvision a dedicated test environment with isolated compute resources and integrate it with the CI/CD pipeline.
   • CContinue using the shared environment but run tests at off-peak hours.
   • DRun tests manually on a local machine and report results in a spreadsheet.
   Reveal Solution Discussion Select an answer first
4. 23expert · hard

   A robotics company is testing a navigation model for a warehouse robot. Real-world testing is expensive and risky. They need to test the model in a variety of scenarios, including edge cases like crowded aisles and sensor failures. They also need to ensure that the test results are consistent across runs. What is the best approach?

   • AUse real-world testing for all scenarios to get the most accurate results.
   • BTest only in a controlled indoor environment with no obstacles.
   • CUse a high-fidelity simulation environment with synthetic scenarios and fixed random seeds for reproducibility.
   • DUse a simulation environment but vary the scenarios randomly each run.
   Reveal Solution Discussion Select an answer first
5. 24application · medium

   An autonomous vehicle company needs to test its perception model in scenarios that are rare in real-world data, such as a pedestrian suddenly crossing in heavy fog. They want to evaluate the model's response without risking safety. What is the best approach?

   • AUse only historical data from other companies without simulation.
   • BTest the model in a real vehicle on public roads during foggy conditions.
   • CUse a simulation environment that generates synthetic fog and pedestrian scenarios to test the model.
   • DWait until enough real-world fog and pedestrian data is collected.
   Reveal Solution Discussion Select an answer first
6. 25foundation · easy

   What does version control of code, data, and models enable in AI testing?

   • AReproducibility of experiments and traceability of changes
   • BElimination of the need for testing
   • CAutomatic model deployment to production
   • DGuarantee of model accuracy
   Reveal Solution Discussion Select an answer first