Selecting Test Techniques for AI-Based Systems CT-AI Practice Questions (Page 1)

Questions 1–5

1. 1foundation · easy

   When selecting test techniques for an AI-based system, which of the following is a key consideration?

   • AThe system's context of use and potential impact of errors
   • BThe brand of the hardware used for training
   • CThe physical size of the training dataset
   • DThe age of the development team
   Reveal Solution Discussion Select an answer first
2. 2expert · hard

   A team is testing a speech recognition system that is non-deterministic due to random sampling in the model. They need to ensure the system is reliable. Which testing approach is most appropriate?

   • AUse a single run of each test case to save time
   • BApply deterministic white-box testing to the model's internal state
   • CRun the same test cases multiple times and analyze the distribution of outputs
   • DUse metamorphic testing with a single input transformation
   Reveal Solution Discussion Select an answer first
4. 3foundation · easy

   Which technique is commonly used to evaluate the performance of a classification AI model?

   • AUsability testing
   • BLoad testing
   • CCode coverage analysis
   • DConfusion matrix
   Reveal Solution Discussion Select an answer first
5. 4application · medium

   A team is testing a chatbot that provides customer support. The chatbot uses a large language model and can generate unexpected responses. The team wants to test the chatbot's behavior with a wide range of user inputs. Which technique is most appropriate?

   • AMetamorphic testing with paraphrased inputs
   • BDifferential testing against a rule-based chatbot
   • CWhite-box testing of the model's internal layers
   • DMutation testing on the training data
   Reveal Solution Discussion Select an answer first
6. 5application · medium

   A team is testing a recommendation system that uses collaborative filtering. The system is data-dependent and its performance varies with the user-item interaction data. The team wants to evaluate the system's performance across different user segments. Which technique is most appropriate?

   • ASubgroup analysis to evaluate performance for different user segments
   • BWhite-box testing of the model's latent factors
   • COverall accuracy metric
   • DAdversarial testing to find edge cases
   Reveal Solution Discussion Select an answer first