Q58 — AWS AIF-C01 Ch.2

Question 58 of 100 | ← Chapter 2

An accounting firm wants to adopt large language models (LLMs) to automate document review. The firm must act responsibly and avoid potential risks. Which measures should the firm take when developing and deploying LLMs? (Choose two.)

• A. Incorporate fairness metrics in model evaluation.
• B. Adjust the model’s temperature parameter.
• C. Modify training data to reduce bias. ✓
• D. Avoid overfitting on training data.
• E. Apply prompt engineering techniques.

Correct Answer: C. Modify training data to reduce bias.

Explanation

When developing and deploying LLMs, organizations must implement measures to ensure fairness, accuracy, and safety. Incorporating fairness metrics during model evaluation (A) helps detect and mitigate discriminatory outcomes. Modifying training data to reduce bias (C) addresses root causes of unfairness by improving dataset representativeness and balance. While options B, D, and E support performance tuning or robustness, they are not primary safeguards against ethical or societal risks like bias or unfairness.