Task Statement 4.1: Explain the Development of AI Systems That Are Responsible. AIF-C01 Practice Questions (Page 3)

Questions 11–15

1. 11foundation · easy

   A model is too simple and performs poorly on both the training data and the test data. Which issue is the model most likely experiencing?

   • AUnderfitting
   • BDemographic bias
   • COverfitting
   • DHigh variance
   Reveal Solution Discussion Select an answer first
2. 12application · medium

   A data science team has trained a model to detect fraudulent transactions. The model performs poorly on both the training data (78% accuracy) and the validation data (76% accuracy). The team believes the model is too simple to capture the complex patterns in the transaction data. Which pair of concepts best describes this situation, and which tool should the team use to investigate it?

   • AUnderfitting and variance; use Amazon A2I to have humans review the model's predictions on the training data
   • BOverfitting and bias; use Amazon SageMaker Clarify to generate a bias report for the training data
   • COverfitting and variance; use Amazon SageMaker Model Monitor to track the model's performance in production
   • DUnderfitting and bias; use Amazon SageMaker Clarify to analyze the training data for patterns the model is missing
   Reveal Solution Discussion Select an answer first
4. 13application · medium

   A retail company is building a product recommendation model. The training data contains purchase history from the company's loyalty program, which has 70% male and 30% female members. The company wants to ensure the recommendation model serves all customers fairly. Which dataset characteristic is most important to address, and what is the likely effect if it is not addressed?

   • ADiversity; the model may exhibit high variance due to the small number of female customers in the dataset
   • BCurated sources; the model may overfit to the loyalty program's specific customer base
   • CInclusivity; the model may hallucinate product recommendations for customers outside the loyalty program
   • DBalance; the model may underperform for female customers due to insufficient representation in the training data
   Reveal Solution Discussion Select an answer first
5. 14application · medium

   A healthcare company is developing an AI system to triage patient messages and route them to the appropriate department. The training dataset was built from messages received at a single urban clinic over six months. During pilot testing, the system performs well overall but shows noticeably lower accuracy for patients over 65 and for those who communicate in Spanish. Which combination of responsible AI features should the team prioritize addressing before deployment?

   • AFairness and inclusivity, supported by dataset diversity and subgroup performance analysis
   • BBias and variance, supported by cross-validation and regularization
   • CVeracity and transparency, supported by citation tracing and confidence scores
   • DRobustness and safety, supported by adversarial testing and input validation
   Reveal Solution Discussion Select an answer first
6. 15foundation · easy

   A team needs to continuously monitor a deployed model to detect when the distribution of incoming data changes over time, which could affect model accuracy. Which AWS service is designed for this ongoing monitoring task?

   • AAmazon Augmented AI (A2I)
   • BAmazon SageMaker Clarify
   • CAmazon Bedrock Guardrails
   • DAmazon SageMaker Model Monitor
   Reveal Solution Discussion Select an answer first