AI Concerns, Ethical Principles, and Responsible AI CRAGE Practice Questions (Page 4)

Questions 16–20

1. 16expert · hard

   A company is developing an AI chatbot for customer service. The chatbot is trained on historical customer interactions, which include some rude and biased responses. The company wants to launch quickly but also wants to avoid reputational damage. Which approach best balances speed and responsible AI?

   • ADelay the launch until the training data is completely cleaned of all biased content.
   • BLimit the chatbot to only answer pre-approved questions and use human agents for everything else.
   • CLaunch the chatbot without any filters to gather real-world data quickly.
   • DImplement a content filter and a human escalation path for sensitive topics, then launch and monitor.
   Reveal Solution Discussion Select an answer first
2. 17expert · hard

   A company develops an AI system that predicts employee turnover. The system uses data including performance reviews and salary. The company wants to use the system to identify employees at risk of leaving and offer retention incentives. The HR department is concerned that the system may be biased against certain groups. The company must decide how to proceed. Which approach best balances the business goal and the fairness concern?

   • AValidate the model's predictions for different demographic groups and adjust the retention program if disparities are found.
   • BOffer retention incentives to all employees to avoid any bias.
   • CDelay the program until the model is perfect and free of any bias.
   • DUse the system to identify at-risk employees but do not disclose the use of AI to employees.
   Reveal Solution Discussion Select an answer first
4. 18expert · hard

   A healthcare provider is deploying an AI diagnostic tool. The tool is very accurate overall but is less accurate for patients with rare diseases. The provider wants to ensure responsible AI deployment. Which action is most aligned with responsible AI?

   • ADeploy the tool and rely on its overall accuracy, as it is above 95%.
   • BDeploy the tool with a clear indication of its confidence level and require a specialist review for low-confidence cases.
   • CUse the tool as the sole decision-maker for all cases to ensure consistency.
   • DLimit the tool to only common diseases and do not use it for rare diseases.
   Reveal Solution Discussion Select an answer first
5. 19application · medium

   A tech company develops a facial recognition system for use in law enforcement. Civil rights groups raise concerns about potential misuse and surveillance. Which ethical concern is most directly relevant to this scenario?

   • AEconomic displacement, because the system may replace human security guards.
   • BSocietal impact, because the technology could enable mass surveillance and infringe on civil liberties.
   • CLack of transparency, because the system's algorithms are proprietary.
   • DTechnical robustness, because the system may fail in low-light conditions.
   Reveal Solution Discussion Select an answer first
6. 20expert · hard

   A multinational corporation is deploying an AI HR system across several countries. The system screens resumes and ranks candidates. The legal team warns that the system may violate local anti-discrimination laws in some countries. The company wants a single global system to reduce costs. Which approach best reconciles the conflicting requirements?

   • ADeploy the system only in countries where it is currently compliant and skip other markets.
   • BDeploy the same system globally and rely on a disclaimer that the company is not liable for local law violations.
   • CUse a separate AI system for each country to ensure full compliance.
   • DDeploy the system with region-specific configurations that adjust features and thresholds to comply with local laws.
   Reveal Solution Discussion Select an answer first