How should organizations approach fairness and bias mitigation in AI systems?

Correct answer: Continuously monitor predictions for bias, use diverse training data, implement fairness metrics, and validate decisions across demographic groups

Option D is correct because responsible AI fairness requires a continuous, multi-layered approach: using diverse and representative training data reduces historical bias, fairness metrics quantify disparate impact across groups, and ongoing monitoring catches bias that emerges after deployment due to data drift or feedback loops. Option A is wrong because sacrificing fairness for raw accuracy is a violation of responsible AI principles and can cause real harm to under-represented groups. Option B is wrong because limiting AI to non-critical applications is neither practical nor a valid mitigation strategy for bias. Option C is wrong because applying identical model parameters to all groups does not account for structural differences in data and can perpetuate or amplify existing disparities.