Fundamentals of Generative AI →hardMultiple Select

Generative AI Leader Fundamentals of Generative AI Practice Question

A product team is designing a generative AI application to create personalized email campaigns. They want to ensure the model produces high-quality, relevant content while minimizing risks such as bias and harmful outputs. Which two practices should the team implement? (Choose two.)

⚠ Common exam trap

The trap here is focusing only on creative output and overlooking the need for systematic bias detection and automated safety filters.

Answer choices

Why each option matters

Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.

Correct answer & explanation

✓

Implement content filters and safety thresholds to block harmful or inappropriate outputs.

Regular fairness evaluations help detect and reduce bias across demographic groups, ensuring equitable and appropriate content. Content filters and safety thresholds act as a safeguard to block harmful or inappropriate outputs before they reach users. Together, these practices address both bias and harmful content, aligning with responsible AI development. Other options either increase randomness, risk overfitting, or remove necessary oversight.

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

✗
Fine-tune the model on a small dataset of previously successful email campaigns without any validation.
Why it's wrong here
Fine-tuning on a small, unvalidated dataset can lead to overfitting and perpetuate existing biases or errors. Without validation, the model may learn undesirable patterns. This approach does not ensure quality or safety; it could amplify problems. Proper fine-tuning requires large, diverse, and carefully vetted datasets, along with evaluation.
✓
Implement content filters and safety thresholds to block harmful or inappropriate outputs.
Why this is correct
Content filters and safety thresholds automatically detect and suppress toxic, biased, or otherwise harmful text before it reaches recipients. This is a critical layer of defense in generative AI applications, especially for customer-facing campaigns. It helps maintain brand safety and compliance with regulations, directly addressing the goal of minimizing harmful outputs.
✓
Conduct regular fairness evaluations of the model's outputs across different demographic groups.
Why this is correct
Fairness evaluations help identify and mitigate biases that could lead to discriminatory or inappropriate content in email campaigns. By testing outputs across demographics, the team can detect disparate performance and adjust prompts, fine-tuning, or filters. This practice aligns with responsible AI principles and reduces reputational and legal risks associated with biased marketing.
✗
Disable all logging and monitoring to protect user privacy and reduce overhead.
Why it's wrong here
Disabling logging and monitoring removes the ability to detect and respond to harmful outputs or model degradation. While privacy is important, logging can be done in a compliant manner with anonymization. Monitoring is essential for ongoing safety and quality assurance. This option would increase risk by eliminating oversight.
✗
Set the model's temperature to the maximum value to encourage diverse and creative emails.
Why it's wrong here
A very high temperature increases randomness, which can lead to incoherent, off-brand, or even harmful content. While creativity is desirable, maximizing temperature sacrifices quality and control. The team should instead tune temperature moderately and combine with other safeguards. This option would likely increase risks rather than mitigate them.

Go deeper

Generative AI Leader Fundamentals of Generative AI Practice Question

Why each option matters

Option-by-option breakdown

Related to this question