Generative AI Leader Practice Test 26 — 15 Questions

Question 1

A software development team builds an internal code assistant using a generative model. The assistant writes Python functions that often contain security vulnerabilities such as SQL injection or command injection. The team wants to mitigate these vulnerabilities without adding a manual review step for every code snippet, as that would slow development. They have access to a static analysis security scanner API. Which approach best addresses the vulnerabilities while maintaining developer velocity?

Choose an answer to begin — your selection is scored in the full session.

Accepted Answer

After each generation, automatically run the code through the static analysis scanner, and if vulnerabilities are found, send the output back to the model for revision with the scanner's feedback.. It creates an automated feedback loop: the static analysis scanner detects vulnerabilities in the generated code, and the model revises the output based on that feedback. This approach directly mitigates security flaws without requiring manual review, preserving developer velocity. It leverages the scanner's precise, rule-based detection to iteratively improve the model's output, which is more reliable than relying on the model's inherent safety.

Answer

Increase top-k sampling to generate a wider variety of code tokens.

Answer

Fine-tune the model on a corpus of secure code examples.

Answer

Add a system prompt: 'Do not generate code with security vulnerabilities.'

Question 2

Which TWO techniques are most effective for improving the quality of a generative AI model's output when summarizing complex documents?

Accepted Answer

Providing few-shot examples of ideal summaries. Option A is correct because few-shot prompting supplies the model with concrete examples of the desired summary format, style, and level of detail, which conditions the model to reproduce that structure and improves fidelity when summarizing complex documents. Option B is correct because a larger, more capable model such as PaLM 2 has greater capacity to comprehend long, intricate documents and generate more accurate, coherent summaries than a smaller predecessor like PaLM. Option C is not appropriate because increasing max output length only allows longer responses; it does not improve summary quality and can even encourage verbosity rather than conciseness. Option D is not appropriate because setting top_p to 0.1 sharply narrows nucleus sampling, reducing diversity and making the output more rigid without addressing summarization quality. Option E is not appropriate because raising temperature to 0.8 increases randomness, which tends to introduce inaccuracies and inconsistency rather than improve the quality of document summaries.

Answer

Increasing max output length significantly

Answer

Setting top_p to 0.1

Answer

Adjusting temperature to 0.8

Question 3

A healthcare company needs to process medical records (e.g., discharge summaries) to extract structured data. Which AI API is specifically designed for this purpose?

Accepted Answer

Document AI with Healthcare NLP. Document AI with Healthcare NLP is purpose-built to parse healthcare documents such as discharge summaries, clinical notes, and medical records, extracting structured entities like medications, diagnoses, and procedures using healthcare-specific models. It is the only option designed specifically for medical record processing.

Answer

Natural Language AI

Answer

Translation AI

Answer

Vision AI

Question 4

A developer is using Vertex AI PaLM 2 to generate product descriptions. The output is often too verbose and includes irrelevant details. Which technique should the developer apply?

Accepted Answer

Use few-shot prompting with examples of concise descriptions. The developer needs to constrain the model's output to be concise and relevant. Few-shot prompting provides the model with explicit examples of the desired output format (concise descriptions), guiding it to mimic that style and length. This directly addresses verbosity and irrelevant details without altering the model's fundamental randomness or safety settings.

Answer

Set top_p to 0.1

Answer

Enable safety filters

Answer

Increase temperature to 0.9

Question 5

A company is using Vertex AI to generate customer support summaries from chat logs. They notice that the summaries sometimes include irrelevant details from the conversation. Which technique should they use to reduce irrelevant details?

Accepted Answer

Add a system instruction to focus on key points.. Adding a system instruction to focus on key points is the most direct and effective technique for reducing irrelevant details in generated summaries. System instructions act as a persistent, high-level directive that guides the model's attention and output structure without altering the underlying model weights. This allows the model to filter out extraneous information from the chat logs by explicitly prioritizing key points, which is a standard practice in prompt engineering for Vertex AI.

Answer

Use a higher top-k value.

Answer

Fine-tune the model on a large dataset of general conversations.

Answer

Increase the temperature parameter.

Question 6

A team is using a pre-trained language model to summarize legal documents. They find that summaries often miss key dates and parties involved. Which technique would most effectively improve factual accuracy?

Accepted Answer

Fine-tune the model on a dataset of legal summaries with annotated key entities.. Fine-tuning on a dataset of legal summaries with annotated key entities directly teaches the model to recognize and reproduce critical factual elements like dates and parties. This supervised learning approach adjusts the model's weights to prioritize entity extraction and accurate generation, which is the most effective method for improving factual accuracy in domain-specific tasks.

Answer

Use top-p sampling with a low p value.

Answer

Increase the temperature parameter.

Answer

Use chain-of-thought prompting.

Question 7

A developer is using the Gemini API to generate marketing copy. They want the output to be diverse and creative but still relevant to the topic. Which THREE parameter adjustments would help achieve this? (Choose 3)

Accepted Answer

Increase temperature to 0.9. Increasing temperature to 0.9 (option A) is correct because higher temperature flattens the probability distribution, making the model more likely to pick less probable tokens and thus produce more diverse, creative text while still grounded in the prompt. Increasing top-k to 50 (option B) is correct because a larger top-k widens the candidate pool from which the next token is sampled, allowing more varied word choices and greater output diversity. Increasing top-p to 0.95 (option E) is correct because a higher nucleus sampling threshold includes more tokens cumulatively comprising 95% of the probability mass, which increases randomness and creativity while excluding only the least likely tokens. Decreasing temperature to 0.1 (option C) is not correct because low temperature makes the model nearly deterministic and conservative, reducing diversity. Decreasing top-k to 10 (option D) is not correct because a smaller top-k restricts sampling to fewer high-probability tokens, which narrows creativity rather than enhancing it.

Answer

Decrease temperature to 0.1

Answer

Decrease top-k to 10

Question 8

A company uses Vertex AI Agent Builder to create a customer support agent. They need the agent to answer questions about order status by calling an internal API. Which Vertex AI feature should they use?

Accepted Answer

Vertex AI Extensions. Extensions in Vertex AI Agent Builder allow the agent to call external APIs (including internal ones) as tools during conversation.

Answer

Vertex AI RAG Engine

Answer

Grounding with Google Search

Answer

Vertex AI Model Garden

Question 9

Which THREE steps are required to secure a generative AI pipeline that uses Vertex AI and involves sensitive customer data?

Accepted Answer

Use VPC Service Controls to create a perimeter around Vertex AI resources. Option A is correct because VPC Service Controls lets you define a service perimeter around Vertex AI resources, preventing data exfiltration of sensitive customer data even if credentials are compromised. Option B is correct because applying IAM roles with least privilege and using dedicated service accounts for the pipeline enforces fine-grained access control and limits the blast radius of any compromised identity. Option D is correct because enabling data encryption at rest with Cloud KMS (customer-managed encryption keys) protects sensitive customer data stored in Vertex AI datasets, models, and related storage from unauthorized access at the storage layer. Option C is incorrect because exposing the prediction endpoint publicly with only an API key removes network-level protections and is not a recommended security control for sensitive data. Option E is incorrect because disabling audit logging reduces visibility and accountability, which weakens security and compliance rather than strengthening them.

Answer

Expose the prediction endpoint publicly with an API key

Answer

Disable audit logging to reduce data exposure

Question 10

A company wants to extract structured data from scanned invoices. Which Google Cloud AI service is purpose-built for this use case?

Accepted Answer

Document AI. Document AI is Google Cloud's purpose-built service for extracting structured data from documents such as invoices, receipts, forms, and contracts. It offers specialized parsers (Invoice Parser, Expense Parser, Form Parser) that return structured fields like invoice number, vendor, line items, and totals, making it the correct choice for scanned invoice extraction.

Answer

Vision AI

Answer

Translation API

Answer

Natural Language AI

Question 11

A company wants to ensure only authorized users can deploy gen AI models. The current policy allows all users in the domain. What is the best practice to restrict deployment?

Accepted Answer

Add condition to restrict deployment. Adding a condition to restrict deployment (e.g., using IAM conditions in Google Cloud's attribute-based access control) allows you to limit model deployment to only authorized users based on attributes like user role, project, or resource tags. This is the best practice because it enforces fine-grained access control without removing existing permissions or adding unnecessary roles, directly addressing the requirement to restrict deployment while maintaining existing user access.

Answer

Remove the binding

Answer

Add more roles

Answer

Use organizational policies

Question 12

A company is deploying a generative AI application that generates medical reports. They need to ensure the output is factual and minimizes hallucinations. Which approach is most effective?

Accepted Answer

Implement retrieval-augmented generation (RAG) with a curated knowledge base. Retrieval-Augmented Generation (RAG) is the most effective approach because it grounds the model's output in a curated, authoritative knowledge base of medical data. By retrieving relevant, verified documents at inference time, RAG directly reduces the model's reliance on its parametric memory, which is the primary source of hallucinations in generative AI. This is especially critical in high-stakes domains like medical reporting, where factual accuracy is paramount.

Answer

Fine-tune the model with RLHF

Answer

Set the temperature to 0.0

Answer

Use prompt engineering to instruct the model to be accurate

Question 13

Which TWO techniques are commonly used to control the style and tone of a generative model's output?

Accepted Answer

Fine-tuning on a dataset with desired style. Fine-tuning on a dataset that embodies the desired style directly adjusts the model's weights, making it consistently produce outputs with that specific tone and style. This is a fundamental technique for customizing generative models, as it teaches the model the exact patterns, vocabulary, and stylistic nuances present in the training data.

Answer

Adjusting the temperature

Answer

Modifying the top_k value

Answer

Changing the top_p value

Question 14

Which THREE components are core to a typical Retrieval Augmented Generation (RAG) system?

Accepted Answer

Vector database. In a typical RAG system, the vector database (B) is core because it stores embedded representations of documents and enables efficient similarity search to retrieve relevant context for a query. The embedding model (C) is also essential, as it converts both the source documents and the user query into dense vector representations that allow semantic matching. The large language model (E) is the third core component, since it consumes the retrieved context along with the query to generate a grounded, natural-language answer. By contrast, a classifier (A) and a rewriter (D) are optional add-ons sometimes used for query routing or query reformulation, but they are not fundamental building blocks of a standard RAG pipeline.

Answer

Classifier

Answer

Rewriter

Question 15

Which THREE factors should you consider when selecting a foundation model from Model Garden? (Choose three.)

Accepted Answer

Model size. When selecting a foundation model from Model Garden, model size (C) matters because parameter count directly affects inference latency, throughput, memory footprint, and cost, so it must match your deployment and budget constraints. Model accuracy on benchmarks (D) is essential because benchmark results (e.g., MMLU, GSM8K, HELM scores) indicate how well the model performs on the tasks relevant to your use case. Model license (E) must be evaluated because it determines whether commercial use, redistribution, or fine-tuning is permitted, which directly impacts legal and business viability. The number of model versions (A) is not a primary selection factor since version count alone says nothing about capability, cost, or fit. The color of the model card (B) is purely cosmetic and has no bearing on model selection.

Answer

Number of model versions

Answer

The color of the model card