NCA-GENL Question 226
Select 3A company is developing a generative AI model to assist with hiring decisions by generating candidate summaries. During testing, the team notices that the model exhibits gender bias in its summaries. Which steps should the company take to minimize bias in their AI system?
- A
Ensure the training data is diverse and representative of all genders and demographics.
- B
Use bias detection tools to evaluate the model outputs and retrain the model if necessary.
- C
Increase the size of the training dataset without modifying its content.
- D
Incorporate fairness-aware algorithms during model development.
- E
Ignore the bias during testing and focus on improving the model's accuracy instead.
Show answer and explanation
Correct answers: A, B, D
Explanation
Minimizing bias in AI systems requires addressing both the data and model development processes. Diverse and representative training data ensures that the model learns equitably across all groups, while fairness-aware algorithms and bias detection tools help mitigate bias during and after training. Simply increasing dataset size or ignoring bias does not resolve the issue.
- A. Correct.
Ensuring diverse and representative training data is a critical step in reducing bias, as biased datasets often lead to biased models.
- B. Correct.
Using tools to detect and mitigate bias in model outputs allows the team to identify issues and adjust the model accordingly.
- C. Incorrect.
Increasing the size of the dataset without addressing its content does not guarantee a reduction in bias; biased data will still lead to biased models.
- D. Correct.
Fairness-aware algorithms are specifically designed to minimize bias and promote equitable outcomes in AI systems.
- E. Incorrect.
Ignoring bias and focusing only on accuracy is counterproductive and could perpetuate unfair outcomes, especially in sensitive applications like hiring.