Generative AI Leader Google Cloud's Generative AI Offerings Practice Test 3 — 15 Questions

Question 1

A company is using Vertex AI Model Garden to deploy a foundation model for document summarization. They notice that the model sometimes generates summaries that include factual errors. They want to reduce hallucinations without sacrificing latency. Which approach should they try first?

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

Accepted Answer

Enable Vertex AI Grounding with a curated database of documents. Vertex AI Grounding connects the model to a curated database of documents, allowing it to retrieve and cite factual information in real-time. This directly reduces hallucinations by grounding responses in verified sources without adding significant latency, as the retrieval step is optimized for speed. Other approaches either increase latency (fine-tuning), reduce output quality (temperature increase), or do not address factual accuracy (safety filters).

Answer

Increase the temperature parameter to make the model more confident

Answer

Add more safety filters to block uncertain responses

Answer

Fine-tune the model on a high-quality dataset of correct summaries

Question 2

A large enterprise is deploying a multi-modal generative AI application that processes customer support emails (text) and attached screenshots (images). They need to run inference on over 10,000 requests per minute with strict latency requirements (p99 < 500ms). They have already selected Gemini 1.5 Pro as the model and deployed it on Vertex AI using a GPU-based endpoint with autoscaling. During testing, they observe that the p99 latency spikes to over 2 seconds during peak traffic. The application is stateless and requests are independent. The team has access to Cloud Observability and can modify the deployment configuration. Which course of action should the team take to meet the latency requirements while minimizing cost?

Accepted Answer

Enable Vertex AI Model Caching and deploy the endpoint on a managed instance group with larger GPU nodes (e.g., A100 40GB). Vertex AI Model Caching reduces latency by caching the model's weights in GPU memory, eliminating the need to reload them for each request. Deploying on larger GPU nodes (A100 40GB) provides higher memory bandwidth and compute capacity, which directly addresses the p99 latency spike by ensuring the model can process more requests per second without queueing. This combination minimizes cost because it optimizes existing GPU utilization rather than simply adding more replicas, which would increase cost without fixing the root cause of latency.

Answer

Increase the maximum number of replicas in the autoscaling configuration to handle spikes

Answer

Use preemptible VMs for the endpoint to get priority scheduling

Answer

Switch to a CPU-based n2-standard instance to reduce GPU contention

Question 3

Which THREE of the following are features of Vertex AI Studio (Gen AI Studio)? (Choose 3)

Accepted Answer

Compare responses from different models side-by-side.. Option C is correct because Vertex AI Studio provides a side-by-side comparison view that lets you send the same prompt to multiple foundation models (e.g., Gemini, PaLM) and evaluate their responses concurrently. Option D is correct because Vertex AI Studio offers a visual, no-code tuning workflow where you upload a custom dataset and fine-tune supported foundation models directly from the console. Option E is correct because prompt design and testing is the core purpose of Vertex AI Studio, allowing users to author, iterate, and test prompts against various foundation models. Option A does not belong because safety filter configuration is handled through Vertex AI's safety settings and policy controls rather than being a defining feature of Studio's prompt/tuning workflow. Option B does not belong because deploying custom container images to Vertex AI endpoints is a function of Vertex AI Prediction/Model Registry, not Vertex AI Studio.

Answer

Configure pre-built safety filters for generated content.

Answer

Deploy custom container images to Vertex AI endpoints.

Question 4

Which TWO of the following are best practices for configuring safety settings in Vertex AI generative models? (Choose 2)

Accepted Answer

Adjust safety thresholds based on the specific use case and audience.. Option B is correct because Vertex AI safety settings expose configurable threshold levels (for example BLOCK_LOW_AND_ABOVE, BLOCK_MEDIUM_AND_ABOVE, BLOCK_ONLY_HIGH) per harm category, and best practice is to tune these thresholds to the application's domain, risk tolerance, and target audience rather than applying one blanket value. Option C is correct because programmatically reviewing generated content—via Vertex AI's safety-related APIs and content moderation capabilities—lets you detect and filter harmful outputs at runtime, complementing the model's built-in filters with an application-level safety layer. Option A is wrong because disabling safety filters removes protection against harmful content and is never a recommended practice, even for creative use cases. Option D is wrong because a single uniform configuration across all models ignores differing use cases, audiences, and risk profiles, which is exactly why thresholds should be tailored. Option E is wrong because always using the maximum threshold over-blocks benign content, degrading usefulness and causing false positives rather than reflecting a balanced, use-case-specific configuration.

Answer

Disable safety filters for maximum creativity.

Answer

Apply the same safety settings to all models in the organization.

Answer

Always use the maximum safety threshold to block all potentially harmful content.

Question 5

A financial services company is building a customer-facing chatbot using Vertex AI Gemini API to answer questions about account balances, transactions, and branch locations. The chatbot must adhere to strict data privacy regulations (e.g., GDPR) that prohibit sending personally identifiable information (PII) to the model provider. The architecture uses a retrieval-augmented generation (RAG) approach where customer queries are passed to a Cloud Run service, which queries a BigQuery database for relevant data and then sends the context along with the query to the Gemini API. The team is concerned that the context may contain PII. They want to minimize modifications to the existing architecture. What step should the team take to ensure compliance?

Accepted Answer

Configure the Gemini API to disable data logging and use the Enterprise tier that ensures data stays within Google Cloud's controls.. Configuring the Gemini API to disable data logging and using the Enterprise tier ensures that data stays within Google Cloud's controls, addressing GDPR concerns without major architectural changes. Option A (anonymization pipeline) is possible but would require significant modifications. Option B (rejecting queries with PII) is unreliable and would break functionality. Option C (third-party proxy) adds complexity and potential compliance risks.

Answer

Build a separate anonymization pipeline using Cloud Data Loss Prevention to remove PII before sending context.

Answer

Modify the chatbot to reject any query that might contain PII based on regex patterns.

Answer

Route all requests through a third-party proxy that strips PII before sending to Gemini.

Question 6

A news organization is using Vertex AI Gemini to summarize articles. They observe that the summaries sometimes contain hallucinated facts—specifically, dates and statistics that are not in the original article. The team is using the default temperature and top_p settings. They want to reduce hallucinations without making summaries too repetitive or overly conservative. They also need to keep latency low. Which action should they take?

Accepted Answer

Enable grounding with Google Search to provide factual source context.. Enabling grounding with Google Search is the correct action because it directly addresses the root cause of hallucinations—lack of factual source context—by allowing the model to cross-reference generated content with real-time, authoritative web data. This approach reduces fabricated dates and statistics without requiring changes to temperature or top_p, which could introduce repetition or conservatism, and it maintains low latency by leveraging Google's infrastructure for retrieval rather than model retraining.

Answer

Increase the temperature to 1.0 and lower top_p to 0.1.

Answer

Fine-tune the model on a large dataset of articles and human-written summaries.

Answer

Lower the temperature to 0.0 and increase top_p to 1.0.

Question 7

A small business wants to use Vertex AI to analyze customer reviews and extract sentiment, product mentions, and overall themes. They have a small dataset of 500 reviews in a CSV file. The team is not experienced with machine learning and wants a pre-built solution that requires minimal coding. They want to start quickly and scale later. Which Google Cloud offering should they use?

Accepted Answer

Cloud Natural Language API for pre-trained sentiment and entity extraction.. Cloud Natural Language API provides pre-trained models for sentiment analysis and entity extraction, requiring minimal coding (just API calls) and no ML expertise. This aligns with the business's need for a quick, scalable, pre-built solution for their small dataset of 500 reviews, avoiding the overhead of custom training or complex prompting.

Answer

Vertex AI Workbench to build a custom sentiment analysis model.

Answer

AutoML Natural Language to train a custom model on their data.

Answer

Vertex AI Gemini API with zero-shot prompting.

Question 8

A research lab is using Vertex AI to generate high-resolution medical images (2560x1920) of cell structures using Imagen. They have fine-tuned the model on their own microscope images. The generated images are sharp but often contain repeating patterns (e.g., identical cell arrangements) that are not biologically plausible. The team suspects the model is overfitting to spatial patterns in the training data. They have already tried increasing the training dataset size and augmenting it with rotations and flips. What additional technique should they try within Vertex AI?

Accepted Answer

Add regularization techniques such as dropout layers or data augmentation that randomly crops and blends patches.. The repeating patterns indicate the model is memorizing spatial arrangements rather than learning generalizable features. Adding regularization like dropout layers or data augmentation that randomly crops and blends patches (e.g., CutMix or MixUp) directly reduces overfitting by forcing the model to focus on local, biologically plausible details rather than memorizing entire image layouts. Vertex AI's training pipelines support custom augmentation strategies, making this a practical and targeted fix.

Answer

Switch to a different foundation model like Stable Diffusion.

Answer

Use a larger batch size during fine-tuning.

Answer

Further increase the resolution of training images to 5120x3840.

Question 9

Which command correctly updates the traffic split?

Accepted Answer

gcloud ai endpoints update-traffic-split my-endpoint --region=us-central1 --traffic-split=model-v2=20,model-v1=80. The `gcloud ai endpoints update-traffic-split` command is the dedicated command for modifying traffic splits between deployed models on a Vertex AI endpoint. It uses the `--traffic-split` flag with key-value pairs (model_id=percentage) to assign traffic percentages, ensuring the total sums to 100. This command directly updates the routing configuration without redeploying models.

Answer

gcloud ai endpoints update my-endpoint --region=us-central1 --remove-deployed-model model-v1 --add-deployed-model model-v2 --traffic-split=20

Answer

gcloud ai models update sentiment-model-v2 --traffic-split=20

Answer

gcloud ai endpoints update my-endpoint --region=us-central1 --update-traffic-split=model-v2=20,model-v1=80

Question 10

A retail company has deployed a customer support chatbot using Vertex AI Agent Builder. The chatbot is configured with a knowledge base stored in BigQuery (user manuals) and Cloud Storage (product images). The agent uses a Gemini 1.5 Pro model for response generation. Users report that the chatbot frequently gives incorrect answers and sometimes does not reference the knowledge base at all. Logs show high latency (average response time > 10 seconds) and many responses are generic or hallucinated. The agent's grounding configuration currently uses the default settings. The development team is considering the following actions: A) Switch to a smaller model like Gemini 1.5 Flash to reduce latency. B) Increase the context window of the model to allow more knowledge base content. C) Enable Vertex AI Search for grounding and configure a search aggregation strategy that retrieves relevant documents from the knowledge base. D) Fine-tune the Gemini model with the company's historical chat logs to improve domain-specific responses. Which action should the team take FIRST to address the issues?

Accepted Answer

Enable Vertex AI Search for grounding and configure a search aggregation strategy that retrieves relevant documents from the knowledge base.. The core issue is that the chatbot is not grounding its responses in the provided knowledge base, leading to hallucinations and generic answers. Enabling Vertex AI Search for grounding directly addresses this by forcing the model to retrieve and cite relevant documents from BigQuery and Cloud Storage before generating a response. This is the foundational step to fix the accuracy problem, as no amount of model tuning or context window expansion will help if the model is not consulting the correct data sources.

Answer

Switch to a smaller model like Gemini 1.5 Flash to reduce latency.

Answer

Increase the context window of the model to allow more knowledge base content.

Answer

Fine-tune the Gemini model with the company's historical chat logs to improve domain-specific responses.

Question 11

A media company wants to let employees quickly summarize long internal reports using a conversational interface without writing any code. They need an enterprise-ready assistant that respects Google Cloud access controls and is managed by Google. Which Google Cloud offering should they use?

Accepted Answer

Gemini for Google Workspace. The scenario calls for a no-code, enterprise-managed assistant tied to existing access controls, which is precisely what Gemini for Google Workspace provides inside Docs, Gmail, and Drive. The other services are developer or security tooling: Agent Builder requires building an agent, Model Garden is a model catalog, and Cloud DLP classifies sensitive data rather than generating summaries.

Answer

Cloud Data Loss Prevention

Answer

Vertex AI Agent Builder

Answer

Vertex AI Model Garden

Question 12

A company is evaluating Google Cloud's generative AI offerings to build a custom chatbot. They want to leverage pre-trained foundation models and also need the ability to fine-tune them with their own data. Which two Google Cloud services or features should they consider? (Choose two.)

Accepted Answer

Vertex AI Fine-Tuning. Vertex AI Gemini models provide pre-trained foundation models for building chatbots, and Vertex AI Fine-Tuning allows customization of these models with proprietary data. Together, they enable the company to leverage pre-trained capabilities and adapt them to their specific needs. The other options either lack fine-tuning support or are not focused on foundation models.

Answer

Vertex AI Model Garden

Answer

Google AI Studio

Answer

Dialogflow CX

Question 13

A large enterprise is evaluating Google Cloud's generative AI offerings for a new customer support application. They need to ensure the solution can be customized with their own data, provides enterprise-grade security and compliance, and supports multimodal inputs (text and images). Which two Google Cloud capabilities should they consider? (Choose two.)

Accepted Answer

Vertex AI Model Garden. Vertex AI Gemini models with tuning provide a customizable multimodal foundation that can be fine-tuned with enterprise data, while Vertex AI Model Garden offers access to a range of foundation models, including multimodal ones, under the same enterprise security umbrella. Together, they enable a secure, compliant, and customizable multimodal solution.

Answer

Vertex AI Agent Builder

Answer

Vertex AI Search

Answer

Cloud Vision API

Question 14

A media company wants to ground its Gemini-powered newsroom assistant in its own archive of published articles so answers cite real sources and stay accurate. The team is evaluating Vertex AI grounding options. Which two grounding sources are designed to let Gemini retrieve from the company's own content? (Choose two.)

Accepted Answer

Vertex AI Search data stores containing the company's indexed article archive.. Vertex AI grounding retrieves from supported sources such as Vertex AI Search data stores, whether the content was ingested from files, Drive, or crawled websites. Both data store ingestion of the archive and website ingestion of the public news domain let Gemini answer with citations to the company's own articles. Google Search grounding and raw bucket paths or manual BigQuery lookups do not provide native retrieval over the private archive.

Answer

Grounding with Google Search, which lets the model cite current web results.

Answer

A Cloud Storage bucket path passed as a grounding source parameter in the generateContent request.

Answer

An embedding index stored in BigQuery that is queried manually before each Gemini call.

Question 15

A healthcare company is building a chatbot to answer patient queries based on their medical documents stored in Cloud Storage. They want to minimize latency and ensure data residency in the EU. Which Vertex AI service should they use?

Accepted Answer

Vertex AI Search with document grounding. Vertex AI Search with document grounding is correct because it allows the chatbot to ground responses in the customer's own medical documents stored in Cloud Storage, ensuring low latency through optimized indexing and retrieval, while supporting data residency controls to keep data within the EU. This service is specifically designed for enterprise search and Q&A over private document repositories, making it ideal for healthcare use cases requiring compliance and fast responses.

Answer

Vertex AI Model Garden with fine-tuning

Answer

Vertex AI Agent Builder with web search

Answer

Vertex AI Codey APIs