Generative AI Leader Practice Exam 59 — 20 Questions

Question 1

A healthcare organization is using Vertex AI to build a generative AI application that summarizes patient notes. They need to ensure that the model does not inadvertently generate or expose protected health information (PHI) in its outputs. Which Google Cloud feature should they implement to detect and redact sensitive data in the model's responses?

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

Accepted Answer

Cloud Data Loss Prevention (DLP) API. Cloud DLP API is designed to discover, classify, and protect sensitive data such as PHI. By integrating it into the output pipeline of the generative AI application, the healthcare organization can automatically detect and redact PHI from model responses, ensuring compliance and preventing data leaks.

Answer

Vertex AI Model Monitoring

Answer

VPC Service Controls

Answer

Cloud Identity-Aware Proxy (IAP)

Question 2

A financial services firm is evaluating Google Cloud generative AI offerings to build an internal assistant that answers employee policy questions using the firm's own documents. The firm requires that answers be grounded in those documents rather than the model's general knowledge, and that the assistant cite the source passages it used. Which two capabilities should the firm rely on? (Choose two.)

Accepted Answer

Citations that return the source passages the model used for its answer. Grounding with Vertex AI Search retrieves authoritative passages from the firm's own corpus so answers rest on internal documents, and citations expose the exact passages used, delivering the provenance the firm demands. Tuning does not guarantee traceable grounding, Google Search grounding pulls public content, and higher temperature only adds randomness, so those do not meet the requirements.

Answer

Raising the model temperature so responses vary more across repeated questions

Answer

Tuning a model on the firm's documents so answers are memorized in the model weights

Answer

Grounding with Google Search to pull current public web results into the response

Question 3

A healthcare company wants to use generative AI to summarize patient notes while ensuring compliance with strict data privacy regulations. They plan to use Google Cloud's Vertex AI. Which two features should they implement to protect sensitive data? (Choose two.)

Accepted Answer

VPC Service Controls. Customer-managed encryption keys (CMEK) and VPC Service Controls are both critical for protecting sensitive healthcare data in Vertex AI. CMEK gives the organization control over encryption keys, while VPC Service Controls prevent unauthorized data exfiltration. Together, they help meet regulatory requirements for data privacy and security.

Answer

Storing data in a public Cloud Storage bucket

Answer

Disabling audit logging

Answer

Public IP addressing for Vertex AI endpoints

Question 4

A logistics company wants an AI agent that answers driver questions about delivery procedures, can look up a shipment's live status through an internal REST API, and can escalate to a dispatcher when it cannot resolve an issue. Which Google Cloud offering is purpose-built for assembling this kind of agent?

Accepted Answer

Vertex AI Agent Builder with tools and a connected data store. An agent framework is needed when a solution must converse, ground answers in internal knowledge and invoke external systems through tools. Vertex AI Agent Builder provides those capabilities natively, including data store grounding and tool calls, plus escalation handling. Data pipelines, feature serving and scheduled jobs each solve adjacent problems but cannot assemble an interactive, tool-using agent on their own.

Answer

Cloud Dataflow with a streaming pipeline for shipment events

Answer

Vertex AI Feature Store for serving driver profile attributes

Answer

Cloud Scheduler jobs that poll the shipment API and email updates

Question 5

A logistics company wants to build a generative AI application that answers employee questions using its internal policy documents while keeping the data inside its own Google Cloud project. The team needs enterprise-grade security, access control, and the ability to choose among multiple foundation models. Which offering should they choose?

Accepted Answer

Vertex AI with Gemini models. Vertex AI provides the enterprise controls, model flexibility, and grounding capabilities needed to build a policy-question answering application inside the company's own project. The productivity, prototyping, and vision services either lack the required governance or address entirely different workloads.

Answer

Google AI Studio

Answer

Cloud Vision API

Answer

Gemini for Google Workspace

Question 6

A healthcare technology company is using Vertex AI to build a generative AI assistant that answers patient questions about medications. They must ensure the assistant does not provide medical advice and adheres to safety guidelines. Which Google Cloud feature should they use to enforce these constraints?

Accepted Answer

Vertex AI safety filters and configurable safety settings. Vertex AI safety filters and configurable safety settings allow developers to set thresholds and categories that block harmful or inappropriate content, including medical advice. This is the only option that provides real-time enforcement of safety guidelines. Monitoring, Explainable AI, and Pipelines serve different purposes and cannot enforce content restrictions.

Answer

Vertex AI Explainable AI

Answer

Vertex AI Pipelines

Answer

Vertex AI Model Monitoring

Question 7

A team needs to generate photorealistic product imagery for an e-commerce catalog from text descriptions, and later needs to edit existing photos by removing unwanted objects while preserving the rest of the scene. Which Google Cloud generative media capabilities should they use for these two tasks, respectively?

Accepted Answer

Imagen for text-to-image generation, and Imagen editing capabilities for object removal and inpainting. Text-to-image generation and masked image editing are distinct capabilities, and both are provided within the same generative media family. Using it for generation from product descriptions and for inpainting-based object removal keeps quality and tooling consistent. Analytics platforms, video pipelines and detection-only models each cover adjacent ground but cannot fulfill the complete pair of tasks described.

Answer

Gemini for text-to-image generation, and BigQuery for object removal

Answer

Vertex AI Vision for text-to-image generation, and Imagen for object removal

Answer

Imagen for text-to-image generation, and a custom-trained object detection model for removal

Question 8

A small marketing agency wants to let its non-technical staff draft blog posts using generative AI without writing any code or managing infrastructure. The agency already uses Google Workspace. Which Google Cloud offering should they adopt to meet this need most directly?

Accepted Answer

Gemini for Google Workspace. Gemini for Google Workspace brings generative assistance into the productivity applications the agency already relies on, so non-technical staff can draft and edit content without any coding or infrastructure work. The other services either require AI engineering effort or perform analysis rather than content generation.

Answer

Cloud Natural Language API

Answer

Vertex AI Agent Builder

Answer

Vertex AI Model Garden

Question 9

A bank's risk team must review every AI-generated customer communication for compliance before it is sent. They want to enforce a policy that blocks any message containing prohibited financial advice and routes flagged messages to a human reviewer. Which Vertex AI capability should they configure?

Accepted Answer

A custom classifier deployed on Vertex AI Endpoints and invoked in the generation workflow. The bank needs an inline, policy-specific gate on each generated message. A custom classifier trained on compliant and non-compliant examples, deployed to a Vertex AI Endpoint, can be called in the generation pipeline to score each message; the application then blocks or routes based on that score. Monitoring, standard safety filters, and offline evaluation do not provide real-time domain policy enforcement and routing.

Answer

Vertex AI Model Monitoring with skew and drift detection

Answer

Vertex AI safety filters configured with adjustable harm thresholds

Answer

Vertex AI Model Evaluation with an automatic metric threshold

Question 10

A university research group wants to experiment with Google's Gemini models through a simple web interface, without writing any code or provisioning cloud infrastructure. They need to upload PDFs, ask questions, and iterate on prompts interactively. Which Google Cloud offering should they use?

Accepted Answer

Vertex AI Studio. Vertex AI Studio is the console-based environment for prompting Gemini models, uploading files such as PDFs, adjusting parameters like temperature and token limits, and saving prompts. It requires no coding or infrastructure provisioning, which matches the research group's interactive experimentation needs. Pipelines, Feature Store, and Model Registry serve orchestration, feature management, and model cataloging respectively, none of which provide a prompt playground.

Answer

Vertex AI Model Registry

Answer

Vertex AI Feature Store

Answer

Vertex AI Pipelines

Question 11

A media company wants to generate realistic images for a new marketing campaign. They need a Google Cloud service that can create images from text prompts and offers enterprise-grade controls for content safety and intellectual property. Which service should they use?

Accepted Answer

Vertex AI Imagen. Vertex AI Imagen is specifically built for generating images from text prompts with enterprise-grade safety and IP controls. It is the Google Cloud offering for text-to-image generation, making it the correct choice for a media company that needs realistic images for marketing while ensuring compliance and safety.

Answer

Vertex AI Vector Search

Answer

Cloud Vision API

Answer

Vertex AI Gemini

Question 12

A logistics company wants its operations analysts to ask natural-language questions such as 'Which routes had the most delays last quarter?' and receive answers grounded in data stored in BigQuery, without analysts writing SQL. The team has no plans to build or train custom models. Which Google Cloud offering should they adopt?

Accepted Answer

BigQuery data canvas. The analysts need conversational analytics over data already in BigQuery with no model training, which is exactly what BigQuery data canvas delivers through Gemini-assisted natural-language exploration that generates SQL and visualizations. The other services target model cataloging, pipeline orchestration, or data integration, none of which provide prompt-driven querying of BigQuery tables for non-technical users.

Answer

Vertex AI Pipelines

Answer

Vertex AI Model Garden

Answer

Cloud Data Fusion

Question 13

A logistics company needs a generative AI model that can accept both text and images as input, and produce text output for describing shipping damage. They want to use a Google Cloud model through the Vertex AI API. Which Gemini model capability should they select?

Accepted Answer

Gemini 1.5 Pro with multimodal input support. A multimodal Gemini model such as Gemini 1.5 Pro can take both text and images and return text, directly satisfying the need to describe shipping damage from photos and written notes. The other services either restrict input to text, produce vectors instead of language, or focus on vision classification rather than generative text output.

Answer

Gemini 1.5 Flash with text-only input

Answer

Vertex AI Vision product for image classification

Answer

Vertex AI Embeddings API for text and image vectors

Question 14

A logistics company wants to build a generative AI application that answers employee questions about internal HR policies. The company's policy documents are already stored in Google Drive and must stay synchronized automatically as they are edited. The team has limited ML engineering resources and prefers a managed, low-code path. Which Google Cloud approach should they choose?

Accepted Answer

Create a Vertex AI Search application with a data store connected to the Google Drive folder, then ground Gemini responses on that data store.. Grounding Gemini on a Vertex AI Search data store connected to Google Drive satisfies both constraints: it keeps answers tied to the latest policy documents automatically, and it delivers a managed, low-code experience. The other approaches either require custom ML work, break synchronization, or scale poorly as the document set grows.

Answer

Use the Gemini API with a long system prompt that pastes the full text of every HR policy document into each request.

Answer

Deploy Gemini on a Compute Engine VM with an attached GPU and write a custom retrieval pipeline that polls the Drive API for changes.

Answer

Export the Drive documents to Cloud Storage, then train a custom Gemini model from scratch on those files using Vertex AI Training.

Question 15

A retail bank uses a Gemini model on Vertex AI to answer customer questions about its mortgage products. Testers report that the model sometimes invents interest rates that do not exist in the bank's rate sheet. The team wants the model to ground its answers only in an approved corpus of product documents and cite the passages it used. Which approach should they implement?

Accepted Answer

Configure Vertex AI Search grounding with the bank's document store and enable citations in the generateContent request.. Grounding with Vertex AI Search connects the model to the bank's authoritative document store, so answers are conditioned on retrieved passages rather than on parametric memory, and the grounding metadata enables citations. Sampling controls only shape randomness, fine-tuning bakes in a snapshot of knowledge without provenance, and token limits only change length.

Answer

Lower the temperature parameter to 0 and increase the topK value so the model samples only from the most probable tokens.

Answer

Fine-tune the Gemini model on the bank's historical customer support transcripts so it learns the correct rates.

Answer

Increase the model's output token limit so it has more room to state the correct rate before answering.

Question 16

A media company wants to use Google Cloud generative AI to produce short video clips from text prompts for internal storyboarding. The team needs a managed model that can generate video from text and wants to evaluate it before committing to production. Which two Google Cloud offerings or capabilities should they use? (Choose two.)

Accepted Answer

Veo on Vertex AI. Veo on Vertex AI provides the actual text-to-video generation, while Vertex AI Model Garden gives the team a managed place to discover and evaluate that model before production adoption. Together they cover both the creative capability and the evaluation step. The remaining services address images, documents, or language translation and cannot generate video.

Answer

Cloud Translation API

Answer

Document AI

Answer

Imagen on Vertex AI

Question 17

A company is using a generative AI model to answer customer questions. The answers are often vague and do not use the company's specific terminology. They want to improve the relevance and specificity of the responses. Which technique should they use?

Accepted Answer

Using retrieval-augmented generation (RAG). Retrieval-augmented generation (RAG) enhances a generative model by retrieving relevant information from a knowledge base and including it in the prompt. This grounds the model's responses in factual, company-specific content, improving relevance and ensuring the use of correct terminology. It is efficient because it does not require retraining the model.

Answer

Increasing the model's temperature

Answer

Fine-tuning the model on company documents

Answer

Reducing the max output tokens

Question 18

A product team uses Gemini via the Vertex AI API to draft customer emails. The drafts are accurate but often too long and include unnecessary background. The team wants shorter, more direct outputs while keeping the same model. Which approach should they take?

Accepted Answer

Revise the prompt to explicitly instruct concise, direct language and specify a target length or format.. The most direct and efficient fix for verbose outputs is to change the prompt to request concise, direct language with a specified length or format. Prompt engineering shapes style immediately without retraining or infrastructure changes. Token limits truncate rather than summarize, fine-tuning is overkill for a style tweak, and top-p affects randomness rather than length.

Answer

Fine-tune the model on a dataset of short emails to permanently change its verbosity.

Answer

Set the max output tokens parameter to a very low value and leave the prompt unchanged.

Answer

Increase the top-p value so the model selects from a narrower set of likely tokens.

Question 19

A global retailer wants to build a generative AI application on Google Cloud that answers employee questions about HR policies. The team must ensure the application uses their private policy documents, keeps answers grounded in those documents, and avoids exposing confidential content to unauthorized staff. Which two Google Cloud components should they include in the architecture? (Choose two.)

Accepted Answer

Vertex AI Search configured with the HR policy documents as a data store. A grounded HR assistant needs a retrieval layer and a generation layer. Vertex AI Search indexes the private policy documents, retrieves relevant passages per question, and supports access controls, while Gemini models on Vertex AI generate answers from those passages with grounding. Feature Store, Pipelines, and Model Registry address structured features, workflow orchestration, and model versioning, none of which provide document grounding or content access control.

Answer

Vertex AI Pipelines for nightly retraining of the Gemini model on HR documents

Answer

Vertex AI Model Registry for versioning the HR policy document set

Answer

Vertex AI Feature Store for storing employee tenure and department features

Question 20

A small startup wants to add image generation to its design tool. The developers want to call a Google Cloud generative AI model through a simple API without provisioning infrastructure, managing model servers, or handling GPU capacity planning. Which Google Cloud offering best fits this requirement?

Accepted Answer

The Imagen model on Vertex AI accessed through the Vertex AI API, using the managed model without deploying a custom endpoint.. Imagen on Vertex AI provides managed, API-based image generation where Google handles the serving infrastructure and scaling. The startup only needs to authenticate and send prompts, which satisfies the no-infrastructure and no-capacity-planning constraints. Deploying models to endpoints or self-hosting on Kubernetes or functions adds operational burden the team explicitly wants to avoid.

Answer

Cloud Run functions that call a locally hosted diffusion model packaged in the function's container image.

Answer

A custom container running Stable Diffusion on Google Kubernetes Engine with node auto-provisioning enabled.

Answer

Vertex AI Model Garden, where they deploy Imagen to a dedicated endpoint with autoscaling replicas.