NCA-GENL Question 208
Select 2A company is deploying an AI-powered customer support chatbot using a large language model (LLM). During testing, the chatbot generates responses that are biased against certain demographics. Which of the following ethical principles of trustworthy AI should the company prioritize addressing in order to resolve this issue?
- A
Fairness and non-discrimination
- B
Transparency and explainability
- C
Robustness and safety
- D
Privacy and data protection
- E
Accountability and governance
Show answer and explanation
Correct answers: A, E
Explanation
To address biased responses in an AI system, the company must focus on fairness and non-discrimination to ensure the system produces equitable outcomes, regardless of demographics. Additionally, accountability and governance are crucial to ensure that processes are in place to monitor, identify, and correct such ethical issues effectively.
- A. Correct.
Fairness and non-discrimination focuses on ensuring that AI systems do not produce biased or unfair outcomes, making it directly relevant to addressing biased responses in the chatbot.
- B. Incorrect.
Transparency and explainability involve understanding how an AI system reaches its conclusions, but this principle alone does not directly address the issue of demographic bias in responses.
- C. Incorrect.
Robustness and safety relate to ensuring the AI system behaves reliably under various conditions, but it does not specifically address ethical concerns like bias.
- D. Incorrect.
Privacy and data protection concern safeguarding user data, which is important but unrelated to resolving bias in the chatbot's responses.
- E. Correct.
Accountability and governance ensure that entities deploying AI systems take responsibility for their behavior, including mitigating biases and ensuring fairness in outcomes.