Ethical, Societal, and Security Concerns in AI AIE Practice Questions (Page 2)

Questions 6–10

1. 6foundation · easy

   Which ethical principle in AI is primarily concerned with ensuring that AI systems do not discriminate against individuals or groups based on protected characteristics?

   • AAccountability
   • BTransparency
   • CPrivacy
   • DFairness
   Reveal Solution Discussion Select an answer first
2. 7expert · hard

   A large hospital network uses an AI model to prioritize patients for organ transplants. The model is trained on historical data that may reflect racial and socioeconomic biases. The hospital must comply with strict regulations on fairness and transparency, but also needs to maintain the model's predictive accuracy to save lives. The data science team proposes several approaches. Which approach best balances fairness, transparency, and accuracy?

   • ARetrain the model from scratch using only the most recent data, which is less biased.
   • BKeep the model as is and rely on human review for all decisions.
   • CReplace the model with a simple rule-based system that is fully transparent and fair.
   • DUse a post-hoc explainability tool to interpret the model's decisions and adjust the decision threshold to reduce disparities.
   Reveal Solution Discussion Select an answer first
4. 8application · medium

   A government agency uses an AI system to screen job applicants for public sector positions. The system was trained on historical hiring data that favored candidates from certain universities. A civil rights group files a complaint alleging that the system discriminates against applicants from other universities. The agency wants to defend its use of the system while ensuring compliance with equal opportunity laws. What should the agency do?

   • AConduct a disparate impact analysis and adjust the model to remove any adverse impact on protected groups
   • BHire a third-party auditor to review the system's code and publish the results
   • CRemove the university feature from the model
   • DArgue that the system is objective because it uses historical data
   Reveal Solution Discussion Select an answer first
5. 9application · medium

   A cybersecurity firm develops an AI model that detects malware. The firm sells the model as a service to enterprise clients. A malicious actor obtains access to the model's confidence scores and uses them to craft malware that the model consistently fails to detect. The firm wants to protect its clients without disrupting the service. What should the firm do?

   • ADisable the confidence score feature entirely
   • BRetrain the model daily with new malware samples
   • CIncrease the model's detection threshold to catch more malware
   • DReduce the detail of confidence scores returned to clients and add noise to the outputs
   Reveal Solution Discussion Select an answer first
6. 10application · medium

   A city government plans to deploy an AI system that predicts which neighborhoods are most likely to experience housing code violations, so that inspectors can be prioritized. Community advocates are concerned that the system will disproportionately target low-income and minority neighborhoods, leading to more fines and displacement. The city wants to use the system while addressing these concerns. What should the city do first?

   • ALimit the system to only flagging violations that are visible from public streets
   • BConduct a fairness audit of the model's predictions across demographic groups before deployment
   • CDeploy the system and monitor outcomes for six months, then adjust if needed
   • DPublish the model's source code on the city's open data portal
   Reveal Solution Discussion Select an answer first