AIF-C01 Question 174
Select 3A data science team is building a machine learning model to detect bias in job application recommendations. The team wants to ensure the dataset used is inclusive and diverse. Which actions should the team take to ensure the dataset meets these characteristics?
- A
Include data from candidates of various genders, ethnicities, and age groups.
- B
Use only data from top-performing candidates to ensure quality predictions.
- C
Balance the dataset to have equal representation of different demographic groups.
- D
Curate the dataset from multiple reliable sources to reduce bias.
- E
Exclude data from underrepresented groups to avoid outliers.
Show answer and explanation
Correct answers: A, C, D
Explanation
To ensure a dataset is inclusive and diverse, you should include data from various demographic groups, balance the representation of those groups in the dataset, and curate data from multiple reliable sources. These practices help create a fair and unbiased model that performs well across different populations. Avoid actions that reduce diversity or disproportionately exclude certain groups, as they introduce bias into the model.
- A. Correct.
Including data from diverse genders, ethnicities, and age groups ensures the dataset is representative of the real-world population and promotes inclusivity.
- B. Incorrect.
Using only data from top-performing candidates introduces bias as it excludes a significant portion of the population and does not represent diversity.
- C. Correct.
Balancing the dataset ensures that all demographic groups are equally represented and prevents the model from being biased toward overrepresented groups.
- D. Correct.
Curating data from multiple reliable sources reduces the risk of bias introduced by a single data source and increases dataset diversity.
- E. Incorrect.
Excluding data from underrepresented groups is counterproductive as it reduces diversity and inclusivity, leading to biased and unfair predictions.