Testing Techniques for AI Solutions AAIA Practice Questions (Page 5)

Questions 21–25

1. 21foundation · easy

   What is the goal of user acceptance testing (UAT) for an AI solution?

   • ATo validate that the AI solution meets business requirements and is acceptable to end users in real-world scenarios
   • BTo measure the model's performance under peak load
   • CTo identify security vulnerabilities
   • DTo verify that the model is correctly implemented
   Reveal Solution Discussion Select an answer first
2. 22foundation · easy

   Which testing activity is most likely to involve end users evaluating the AI solution in a production-like environment?

   • AUser acceptance testing
   • BUnit testing
   • CSecurity testing
   • DPerformance testing
   Reveal Solution Discussion Select an answer first
4. 23expert · hard

   A university is developing an AI system to predict student dropout. The available historical data is heavily skewed: 95% of students graduate, and the data for dropouts is sparse and incomplete. The university wants to ensure the model is fair across different socioeconomic backgrounds and that it is acceptable to academic advisors. Which test data preparation strategy best supports these goals?

   • AUse the data as-is and rely on the model's overall accuracy to validate performance.
   • BRemove all dropout records to create a clean dataset of graduates only.
   • COversample the dropout class and impute missing data, then split into training and test sets while preserving the socioeconomic distribution.
   • DCombine training and test data to maximize the amount of data for the minority class.
   Reveal Solution Discussion Select an answer first
5. 24foundation · easy

   What is the purpose of integration testing for an AI solution?

   • ATo measure the model's accuracy on a test set
   • BTo ensure the AI solution integrates correctly with existing systems, data pipelines, and user workflows
   • CTo test the model's ability to handle adversarial inputs
   • DTo evaluate the model's performance under peak load
   Reveal Solution Discussion Select an answer first
6. 25application · medium

   A bank is developing a credit-scoring model. The historical loan data contains 90% approvals and 10% denials, and the denial records are missing several key fields. The model will be used to assess loan applications from a diverse customer base. The data science team plans to train and test the model on this dataset. Which test data preparation approach best supports reliable model validation?

   • AUse the historical data as-is for both training and testing, since it reflects real-world approval rates.
   • BOversample the denial records and impute missing fields, then split the data into separate training and test sets.
   • CRemove all records with missing fields and train only on the approval records to ensure data quality.
   • DCombine training and test data into one set to maximize the amount of data available for model training.
   Reveal Solution Discussion Select an answer first