ISACA Certified Artificial Intelligence Auditor (CAIA) — Question 30

Which of the following controls BEST helps reduce the risk of hallucinations in an AI model?

Answer options

• A. Requiring model output explanations
• B. Implementing a human-in-the-loop approach
• C. Improving model transparency
• D. Reducing the model’s creativity score

Correct answer: D

Explanation

Reducing the model’s creativity score limits its ability to generate novel or unexpected outputs, which can help to diminish the chances of hallucinations. The other options, while beneficial for various aspects of AI deployment, do not directly target the reduction of hallucinations in the same effective manner as adjusting the creativity score.