A human resources team wants to use an LLM to categorize employee feedback into themes like 'workload', 'management', and 'compensation'. They have a dataset of 20 examples per theme. They want to improve accuracy over zero-shot prompting. Which technique should they use?

Use role prompting to have the model act as an HR specialist.

Use zero-shot prompting with a detailed description of each theme.

Use one-shot prompting with one example per theme.

Use few-shot prompting with several examples per theme.

A startup wants to quickly prototype an LLM-powered tool that summarizes news articles. They have no labeled examples and need a fast, cost-effective solution. The summaries just need to be concise and capture the main points. Which prompting technique should they use?

Use role prompting to have the model act as a news editor.

Use prompt chaining to summarize in multiple steps.

Use few-shot prompting with several example summaries.

Use zero-shot prompting with a clear instruction to summarize the article.

Which statement best describes the role of prompt engineering in improving LLM outputs?

It provides clear instructions and context to steer the model's response

It guarantees the model will always produce correct answers

It increases the model's training data size

It adjusts the model's parameters to improve performance

A math tutoring platform wants to use an LLM to solve word problems and show the solution steps to students. The problems involve multiple arithmetic operations and logical reasoning. The team wants the model to produce accurate answers with clear explanations. Which prompting technique should they use?

Use zero-shot prompting and ask for the final answer only.

Use few-shot prompting with examples of solved problems.

Use role prompting to have the model act as a math teacher.

Use chain-of-thought prompting and instruct the model to reason step-by-step.

When is one-shot prompting particularly useful?

When the task is extremely complex and requires multiple reasoning steps

When you want the model to adopt a specific persona

When you need to show the model the exact format for the output

When you want the model to use external tools

Google CloudGenerative AI Leader

Domain 3Objective 2

3.2 Describe Prompt Engineering Techniques and How They Drive Better Results. GENERATIVE-AI-LEADER Practice Questions (Page 2)

Part of the Techniques to improve gen AI model output domain, which accounts for ~20% of the GENERATIVE-AI-LEADER exam.

21questions here

5free pages

9concepts

~20%of the exam

Questions 6–10

6application · medium
A human resources team wants to use an LLM to categorize employee feedback into themes like 'workload', 'management', and 'compensation'. They have a dataset of 20 examples per theme. They want to improve accuracy over zero-shot prompting. Which technique should they use?
Select an answer first
7application · medium
A startup wants to quickly prototype an LLM-powered tool that summarizes news articles. They have no labeled examples and need a fast, cost-effective solution. The summaries just need to be concise and capture the main points. Which prompting technique should they use?
Select an answer first
8foundation · easy
Which statement best describes the role of prompt engineering in improving LLM outputs?
Select an answer first
9application · medium
A math tutoring platform wants to use an LLM to solve word problems and show the solution steps to students. The problems involve multiple arithmetic operations and logical reasoning. The team wants the model to produce accurate answers with clear explanations. Which prompting technique should they use?
Select an answer first
10foundation · easy
When is one-shot prompting particularly useful?
Select an answer first

Finished these 5 questions?

Review the revealed explanations, or continue through the curriculum.

Next 5 questions →Back to GENERATIVE-AI-LEADER

Free Basic Practice is a study aid with revealable answers — not a scored exam. Examers.io is independent and not affiliated with or endorsed by Google Cloud. “GENERATIVE-AI-LEADER” is a trademark of its owner, used for identification only.