Core AI Concepts and Business Applications CAIPM Practice Questions (Page 7)

Questions 31–35

1. 31expert · hard

   A company wants to deploy an AI system to screen job applicants. The system uses a machine learning model trained on resumes of past employees. The HR team is concerned about bias and wants to ensure fair hiring. The legal team warns that the model may inadvertently discriminate. What is the most effective way to address this limitation?

   • AConduct a bias audit on the model's predictions and adjust the training data or model to mitigate identified biases.
   • BUse a more complex deep learning model to improve accuracy and reduce bias.
   • CRemove all demographic data from the training data to ensure fairness.
   • DAbandon the AI system because bias is unavoidable.
   Reveal Solution Discussion Select an answer first
2. 32application · medium

   A logistics company is considering AI to optimize delivery routes. The operations manager expects the AI to reduce fuel costs and delivery times. However, the company's drivers have varying levels of experience and sometimes deviate from suggested routes due to road closures. What is the most realistic value proposition for this AI initiative?

   • AAI will only be valuable if it replaces the existing routing software entirely.
   • BAI will provide route recommendations that reduce costs, but human drivers must remain in the loop to handle exceptions.
   • CAI will guarantee zero delivery delays regardless of external conditions.
   • DAI will eliminate the need for human dispatchers and drivers.
   Reveal Solution Discussion Select an answer first
4. 33application · medium

   A company is evaluating a vendor's AI solution that claims to use 'deep learning' to improve customer churn prediction. The company's data is primarily tabular (customer demographics, usage history). The team is concerned about the complexity of the solution. What is the most accurate assessment of the vendor's claim?

   • ADeep learning is always the best choice for tabular data.
   • BDeep learning is not a form of machine learning.
   • CDeep learning requires no data preprocessing.
   • DDeep learning is a subset of machine learning that may or may not outperform simpler models on tabular data.
   Reveal Solution Discussion Select an answer first
5. 34application · medium

   A retail bank wants to reduce fraudulent credit-card transactions. The fraud team currently uses a rules-based system with manually tuned thresholds. They have years of labeled transaction data and want a solution that can adapt to new fraud patterns without requiring the team to rewrite rules each month. Which approach best meets this requirement?

   • ADeploy a supervised machine learning model trained on historical labeled transactions to score new transactions for fraud likelihood.
   • BExpand the existing rules-based system by adding more if-then conditions for each new fraud pattern.
   • CImplement a simple lookup table of known fraudulent card numbers and block those cards.
   • DUse a deep learning model that automatically learns features from raw transaction sequences without any labeled data.
   Reveal Solution Discussion Select an answer first
6. 35application · medium

   A manufacturing company wants to use AI to predict equipment failures before they occur. They have sensor data from machines, maintenance logs, and historical failure records. Which AI application best fits this scenario?

   • ANatural language processing to read maintenance logs and generate work orders.
   • BPredictive maintenance using machine learning to identify patterns that precede failures.
   • CRobotic process automation to automate the maintenance scheduling process.
   • DComputer vision to visually inspect machines for defects.
   Reveal Solution Discussion Select an answer first