Objective 4.4 Address Business Risks, Ethical Concerns, and Related Concepts in Operationalizing the Model AIP-210 Practice Questions (Page 3)

Questions 11–15

1. 11expert · hard

   A multinational company is deploying a facial recognition model for employee time tracking. The model is trained on a dataset that is predominantly light-skinned individuals. The company operates in several countries with diverse populations. The legal team is concerned about compliance with local privacy laws and potential bias. What is the most appropriate course of action?

   • ADeploy the model only in countries where it is legally permitted and where the employee population matches the training data.
   • BUse the model but add a manual review process for all time-tracking decisions.
   • CObtain explicit consent from all employees before using the model, and proceed with deployment.
   • DConduct a bias assessment on the model using a diverse test set, and if bias is found, retrain with a more representative dataset before deployment.
   Reveal Solution Discussion Select an answer first
2. 12expert · hard

   A logistics company uses an ML model to optimize delivery routes. The model was trained on data from a region with good road infrastructure. The company expands to a new region with poor roads and frequent traffic jams. The model's routing suggestions are often suboptimal, leading to delays and increased fuel costs. The company wants to minimize financial losses while adapting the model. What is the best approach?

   • AUse a hybrid approach: keep the existing model for familiar routes and use manual routing for the new region until enough data is collected.
   • BIncrease the model's learning rate to adapt faster to the new region's data.
   • CDeploy the model in the new region and monitor its performance for a month before making any changes.
   • DImmediately retrain the model on data from the new region, even if the dataset is small.
   Reveal Solution Discussion Select an answer first
4. 13foundation · easy

   A healthcare provider uses an ML model to prioritize patient appointments. Which of the following is an operational business risk of this deployment?

   • AThe model's decisions may be difficult to explain to patients.
   • BThe model may produce biased recommendations against certain patient groups.
   • CThe model may experience downtime, causing delays in patient scheduling.
   • DThe model may violate patient privacy regulations.
   Reveal Solution Discussion Select an answer first
5. 14application · medium

   A media company is deploying a content moderation model. The model sometimes flags benign content as harmful, leading to user complaints. The company's customer support team is overwhelmed with appeals. The data science team wants to improve the model. What should they do to address the ethical and operational concerns?

   • ARemove the model and rely on human moderators only.
   • BPublish the model's moderation rules to users so they understand why content is flagged.
   • CProvide a clear appeals process and use user feedback to retrain the model.
   • DIncrease the model's sensitivity to catch more harmful content, even if it increases false positives.
   Reveal Solution Discussion Select an answer first
6. 15foundation · easy

   An e-commerce company deploys a recommendation model. Which of the following is an ethical risk that should be assessed during deployment?

   • AThe model may require frequent retraining to maintain accuracy.
   • BThe model may recommend products that are out of stock, frustrating customers.
   • CThe model may show biased recommendations that exclude certain product categories.
   • DThe model may increase server costs during peak shopping seasons.
   Reveal Solution Discussion Select an answer first