Principles of Ethical and Fair AI Use AIE Practice Questions (Page 6)

Questions 26–30

1. 26expert · hard

   A company is developing an AI system to screen job applications. The system is designed to reduce hiring time, but it has been shown to rank male candidates higher than female candidates for certain roles, even when qualifications are similar. The company has a legal obligation to avoid discrimination. The hiring team wants to deploy the system quickly. What is the most ethical and legally sound course of action?

   • ADeploy the system immediately and monitor for discrimination after a few months.
   • BRemove gender from the model inputs and deploy, assuming the bias is eliminated.
   • CAdjust the model to remove the gender bias, validate it on a test set, and only then deploy.
   • DUse the system only for initial screening but have human recruiters make the final decision.
   Reveal Solution Discussion Select an answer first
2. 27application · medium

   A company develops an AI chatbot to handle customer service inquiries. The chatbot sometimes provides incorrect information, and customers are frustrated because they cannot understand why the chatbot gave a certain answer. The company wants to improve transparency and explainability. Which action is most effective?

   • AProvide customers with the ability to see the chatbot's confidence level and the sources it used to generate an answer.
   • BHide the chatbot's limitations to maintain customer trust.
   • CUse a more complex language model to generate more accurate responses.
   • DMake the chatbot's responses more verbose to appear more authoritative.
   Reveal Solution Discussion Select an answer first
4. 28expert · hard

   A company is deploying an AI system that makes real-time pricing decisions for its products. The system is complex and not fully interpretable. A consumer advocacy group has raised concerns that the system may be pricing out low-income customers. The company wants to be accountable. What is the most effective way to ensure accountability?

   • APublish the model's source code to the public.
   • BAppoint a single executive to be personally responsible for all pricing decisions.
   • CConduct regular audits of the pricing decisions to assess their impact on different customer segments.
   • DStop using the AI system and revert to manual pricing.
   Reveal Solution Discussion Select an answer first
5. 29application · medium

   A social media company is planning to deploy an AI content moderation system that will automatically remove posts that violate its policies. The system is known to have a higher false-positive rate for posts in certain dialects. The company wants to assess the ethical impact before launch. Which stakeholder group should be included in the impact assessment?

   • AOnly the users who post in the affected dialects, since they are the most directly impacted.
   • BOnly the data scientists who built the model, since they understand its technical limitations.
   • CAll users, content creators, advertisers, and the broader public, with special attention to the affected dialect communities.
   • DOnly the company's legal and compliance teams, since they are responsible for policy enforcement.
   Reveal Solution Discussion Select an answer first
6. 30expert · hard

   A government agency deploys an AI system to determine eligibility for social benefits. The system is developed in-house but uses a third-party algorithm. A citizen appeals a denial, and the agency cannot explain why the decision was made. The agency must ensure accountability and transparency. Which combination of measures is most effective?

   • ADiscontinue the AI system and process all applications manually.
   • BRely on the third-party vendor to provide explanations on request.
   • CImplement a logging system that records all inputs, outputs, and decision rationale, and establish an independent review board to handle appeals.
   • DPublish the algorithm's source code to the public.
   Reveal Solution Discussion Select an answer first