AWS Certified AI Practitioner · Free Practice Question Medium
Question 18
A company is developing an AI model and wants to ensure it does not discriminate against any group. Which two core dimensions of responsible AI should they focus on? (Select two)
-
A
Safety
-
B
Scalability
-
C
Fairness
-
D
Transparency
-
E
Veracity
Reveal correct answers
Correct answers: C, D
Explanation
1. Fairness
Fairness in AI focuses on ensuring that the model treats all groups equitably and does not exhibit bias against specific demographic groups based on factors such as race, gender, age, or socioeconomic status. Addressing fairness involves:
Bias Mitigation: Implementing strategies to detect and reduce bias in training data, model parameters, and outputs.
Equitable Outcomes: Ensuring that the model’s predictions or decisions do not systematically disadvantage or favor any particular group.
By prioritizing fairness, the company can work toward creating a model that provides equitable treatment and minimizes the risk of discriminatory behavior.
2. Transparency
Transparency involves making the AI model’s processes, decisions, and limitations understandable and explainable to users and stakeholders. Focusing on transparency helps in several ways:
Accountability: When the model’s decision-making process is transparent, it becomes easier to identify and address sources of potential bias or unfair treatment.
Explainability: Providing clear explanations of how the model makes decisions can help affected groups and stakeholders understand its impact and ensure that it aligns with ethical standards.
Transparency allows stakeholders to scrutinize the model’s inner workings, which is essential for identifying and correcting any unfair or biased behaviors, contributing to responsible AI development.
Why Other Options Are Less Suitable
Veracity:
Definition: Veracity refers to the accuracy and reliability of the AI model’s outputs.
Irrelevance to Bias: While accuracy is important, it does not directly address fairness or the prevention of discrimination. A highly accurate model can still exhibit bias if fairness considerations are not incorporated.
Safety:
Definition: Safety in AI is about ensuring that the model’s operations and outcomes do not harm users or the environment.
Different Focus: Safety primarily pertains to preventing physical or psychological harm rather than addressing fairness or bias. While it is crucial, it does not specifically help in avoiding discrimination.
Scalability:
Definition: Scalability relates to the model’s ability to perform well across different environments, datasets, or use cases.
Unrelated to Fairness: While a scalable model is beneficial for operational purposes, it does not inherently ensure fairness or prevent discrimination.
References:
AWS Responsible AI: Building AI Responsibly
AWS AI Service Cards: Enhancing Transparency
A.
Safety ensures the model avoids harmful outcomes.
B.
Scalability refers to the ability of the AI system to handle increasing amounts of work or to be expanded across different environments. While important, it is not directly related to the fairness or explainability of the model.
C.
Fairness is a core dimension of responsible AI that ensures the model treats all individuals and groups equally and does not discriminate against any particular group. It involves promoting inclusiveness and addressing biases in the training data and model algorithms.
D.
Transparency ensures that the AI model's decisions and processes are understandable and explainable, which is key to detecting and correcting potential biases.
E.
Veracity focuses on reliability and accuracy.
Discussion
Think the marked answer is wrong, or have a better explanation? Share it below — comments appear after review.
