Question 17
Domain 2: Fairness, Bias, and Societal ImpactA team is designing an AI-driven content recommendation system that may influence users' beliefs or behavior. Which safeguard is most appropriate to reduce the risk of harmful influence while preserving user autonomy?
Correct answer: B
Explanation
When content systems can shape beliefs or behavior, safeguards should be built in to limit harmful influence and support user autonomy. The goal is not only system performance, but also protection against societal harm from recommendations and personalization. — Safeguards for content systems that influence beliefs or behavior.
Why each option is right or wrong
A. Increase personalization intensity so users engage more deeply with recommended content
Safeguards are intended to reduce harmful influence, not strengthen a system's ability to shape behavior.
B. Implement safeguards that limit harmful influence on beliefs or behavior and preserve user autonomy
The source material specifically recommends safeguards for content systems that influence beliefs or behavior. In this scenario, the system is AI-driven and may affect users' beliefs or actions, so the appropriate measure is to add protections that reduce harmful influence while preserving autonomy.
C. Remove all recommendation features because any influence on users is inherently unacceptable
The material calls for safeguards for content systems, not a blanket prohibition on all systems with influence.
D. Focus only on improving prediction accuracy because accurate recommendations prevent societal concerns
Accuracy alone does not address the need for safeguards when systems influence beliefs or behavior.