Testing Techniques for AI Solutions AAIA Practice Questions (Page 1)

Questions 1–5

1. 1application · medium

   A team has updated an AI model that powers a recommendation engine. The update was intended to improve accuracy, but the team is concerned that the change might have introduced new defects or degraded existing functionality. Which testing activity should they perform?

   • AExplainability testing to generate explanations for the model's decisions.
   • BPerformance testing to measure the model's inference speed.
   • CRegression testing to re-run existing test cases and compare results with the previous version.
   • DBias testing to check for demographic disparities.
   Reveal Solution Discussion Select an answer first
2. 2foundation · easy

   What is the primary goal of functional testing for an AI solution?

   • ATo measure the AI solution's response time under peak load
   • BTo verify that the AI solution meets specified functional requirements and produces correct outputs
   • CTo identify security vulnerabilities in the AI solution
   • DTo evaluate the AI solution's ability to handle invalid inputs
   Reveal Solution Discussion Select an answer first
4. 3application · medium

   An insurance company has developed an AI system that automates claim approvals. Before full deployment, the company wants to ensure that claims adjusters find the system's decisions acceptable and that the system meets business requirements. Which testing activity is most appropriate?

   • ASecurity testing to protect against fraudulent claims.
   • BPerformance testing to ensure the system can handle a high volume of claims.
   • CRobustness testing with malformed claim submissions.
   • DUser acceptance testing with a group of claims adjusters using the system on real-world scenarios.
   Reveal Solution Discussion Select an answer first
5. 4foundation · easy

   What is a key characteristic of a well-prepared test data set for validating AI model performance?

   • AIt includes only adversarial examples to stress-test the model
   • BIt is representative, unbiased, and sufficient to cover the expected input space
   • CIt contains only the most common examples to ensure the model performs well on typical cases
   • DIt is identical to the training data to ensure consistency
   Reveal Solution Discussion Select an answer first
6. 5foundation · easy

   Which of the following is an example of continuous testing in an AI operations context?

   • AAutomatically re-evaluating model accuracy on a rolling basis as new data arrives
   • BConducting a single security audit at the end of development
   • CPerforming a one-time functional test before deployment
   • DRunning a user acceptance test with a small group of users
   Reveal Solution Discussion Select an answer first