Bias auditing and fairness assessment: Practice Questions — Safety, Ethics and Compliance (NVIDIA-Certified Professional: Generative AI LLMs)

Bias Auditing and Fairness Assessment Practice Questions

These multiple-choice questions are designed to help candidates prepare for the Safety, Ethics and Compliance section of the NVIDIA-Certified Professional: Generative AI LLMs exam, focusing specifically on bias auditing and fairness assessment.

1. Which of the following best describes the primary goal of bias auditing in large language models?

   • A. To improve model accuracy on training data
   • B. To identify and measure unfair or prejudiced outputs
   • C. To optimize model inference speed
   • D. To reduce the model size for deployment

   Correct Answer: B

   Explanation: Bias auditing aims to detect and quantify unfair or prejudiced outputs generated by the model, ensuring fairness and ethical compliance.

2. Which metric is commonly used to assess fairness in classification outputs during bias auditing?

   • A. Precision-Recall Curve
   • B. Equal Opportunity Difference
   • C. BLEU Score
   • D. Latency

   Correct Answer: B

   Explanation: Equal Opportunity Difference measures disparities in true positive rates between demographic groups, a key fairness metric.

3. What is a key step in mitigating bias after detecting it through auditing?

   • A. Increasing the model’s parameter count
   • B. Applying data augmentation or reweighting techniques
   • C. Removing all rare words from the training data
   • D. Reducing the batch size during training

   Correct Answer: B

   Explanation: Data augmentation or reweighting helps balance representation in training data, reducing biased behavior in model outputs.

4. During fairness assessment, why is it important to evaluate model outputs across multiple demographic groups?

   • A. To improve overall model speed
   • B. To ensure the model performs equitably and does not disadvantage any group
   • C. To increase the size of the training dataset
   • D. To reduce GPU memory usage

   Correct Answer: B

   Explanation: Evaluating multiple groups ensures the model’s fairness and prevents systemic discrimination against any demographic.

5. Which of the following is an example of a guardrail implemented to reduce bias in generative AI models?

   • A. Limiting the maximum token length
   • B. Filtering outputs containing harmful or biased language
   • C. Increasing the learning rate
   • D. Using dropout layers during training

   Correct Answer: B

   Explanation: Guardrails like output filtering help prevent the generation of biased or harmful content, enhancing model safety.

6. What role does human-in-the-loop (HITL) play in bias auditing and fairness assessment?

   • A. Automates all bias detection without human intervention
   • B. Provides qualitative review and contextual judgment to complement automated audits
   • C. Speeds up model training
   • D. Reduces the need for diverse datasets

   Correct Answer: B

   Explanation: HITL allows experts to review outputs for subtle biases that automated tools may miss, improving audit quality.

7. Which approach is most effective for continuous bias monitoring in deployed generative AI systems?

   • A. One-time bias audit before deployment
   • B. Periodic re-evaluation using real-world user data and feedback
   • C. Ignoring bias after initial training
   • D. Only monitoring model latency and throughput

   Correct Answer: B

   Explanation: Continuous monitoring with real-world data helps detect emerging biases and ensures ongoing fairness compliance.

More in this topic

Guardrails implementation: Worked Example — Safety, Ethics and Compliance (NVIDIA-Certified Professional: Generative AI LLMs)Bias auditing and fairness assessment: Common Mistakes — Safety, Ethics and Compliance (NVIDIA-Certified Professional: Generative AI LLMs)Safety, Ethics and Compliance — NVIDIA-Certified Professional: Generative AI LLMsBias detection and mitigation — Safety, Ethics and Compliance (NVIDIA-Certified Professional: Generative AI LLMs)Guardrails implementation: Quick Reference — Safety, Ethics and Compliance (NVIDIA-Certified Professional: Generative AI LLMs)Guardrails implementation: Practice Questions — Safety, Ethics and Compliance (NVIDIA-Certified Professional: Generative AI LLMs)Bias detection and mitigation: Worked Example — Safety, Ethics and Compliance (NVIDIA-Certified Professional: Generative AI LLMs)Bias auditing and fairness assessment: Worked Example — Safety, Ethics and Compliance (NVIDIA-Certified Professional: Generative AI LLMs)Bias detection and mitigation: Practice Questions — Safety, Ethics and Compliance (NVIDIA-Certified Professional: Generative AI LLMs)Bias auditing and fairness assessment: Quick Reference — Safety, Ethics and Compliance (NVIDIA-Certified Professional: Generative AI LLMs)Bias auditing and fairness assessment — Safety, Ethics and Compliance (NVIDIA-Certified Professional: Generative AI LLMs)Guardrails implementation — Safety, Ethics and Compliance (NVIDIA-Certified Professional: Generative AI LLMs)Guardrails implementation: Common Mistakes — Safety, Ethics and Compliance (NVIDIA-Certified Professional: Generative AI LLMs)Bias detection and mitigation: Quick Reference — Safety, Ethics and Compliance (NVIDIA-Certified Professional: Generative AI LLMs)Bias detection and mitigation: Common Mistakes — Safety, Ethics and Compliance (NVIDIA-Certified Professional: Generative AI LLMs)