Generative AI Leader Practice Exam 4 — 20 Questions

Question 1

A company is evaluating Google Cloud's generative AI offerings for enterprise use. Which TWO considerations are most important when selecting the right model deployment option?

Accepted Answer

Data residency. Data residency (A) is a critical consideration because enterprise deployments must comply with regional and regulatory requirements, and Google Cloud's Vertex AI lets you pin model endpoints and data processing to specific locations such as us-central1 or europe-west4 to satisfy sovereignty and compliance mandates. Latency requirements (D) are equally important because the deployment option determines network proximity and serving characteristics — for example, a regional endpoint close to users or a provisioned throughput configuration reduces round-trip time, which directly affects real-time generative AI application performance. Together, residency and latency drive the choice between options like global vs. regional endpoints, on-demand vs. provisioned throughput, and Vertex AI vs. Gemini API. Developer preference (B) is subjective and does not determine the correct deployment architecture for enterprise compliance or performance needs. Model size (C) is a model characteristic rather than a deployment-selection criterion, since Vertex AI abstracts the underlying infrastructure. Training time (E) is irrelevant to deployment selection because it concerns the model-building phase, not how the model is served in production.

Answer

Developer preference

Answer

Model size

Answer

Training time

Question 2

A machine learning engineer is building a text-to-image model using Vertex AI. They want to reduce inference latency. Which strategy is most effective?

Accepted Answer

Use a smaller model variant. Using a smaller model variant directly reduces the number of parameters and computational operations required per inference pass, which lowers latency. In text-to-image models like Imagen or Stable Diffusion, the model size is the primary driver of forward-pass time, so a smaller variant (e.g., fewer layers or reduced latent dimensions) yields faster generation.

Answer

Use a larger image resolution

Answer

Enable batch processing

Answer

Increase the number of inference steps

Question 3

A developer is building a multimodal application that needs to analyze images, understand spoken language, and generate text responses. Which Google Cloud generative AI model is BEST suited for this task?

Accepted Answer

Gemini 1.5 Pro. Gemini models are natively multimodal, accepting text, image, audio, and video inputs. PaLM 2 is text-only. Imagen is for image generation. Chirp is for speech recognition/synthesis.

Answer

PaLM 2 for text generation

Answer

Imagen for image generation

Answer

Chirp for speech recognition

Question 4

During a proof-of-concept for a GenAI document summarization tool, the team wants to evaluate whether the summaries are accurate and retain key information before scaling. Which evaluation approach is most appropriate for this stage?

Accepted Answer

Run an A/B test with a small user group and have domain experts manually review a sample of summaries for accuracy. A/B testing with manual review by domain experts provides qualitative and quantitative feedback on accuracy and completeness, which is crucial for a pilot. Automated metrics alone may not capture business relevance.

Answer

Measure latency and cost as the primary evaluation metrics

Answer

Deploy to all users and collect feedback via a survey

Answer

Use ROUGE scores exclusively to compare summaries against human-written ones

Question 5

Which THREE are best practices for designing prompts for a generative AI model?

Accepted Answer

Provide few-shot examples for complex tasks. Providing few-shot examples (e.g., 2-5 input-output pairs) helps the model infer the desired pattern, reducing ambiguity for complex tasks like classification or structured extraction. This technique leverages in-context learning, where the model uses the examples as a template without fine-tuning.

Answer

Use negative prompts to avoid undesired outputs

Answer

Always set temperature to 1.0 for creativity

Question 6

A company uses Vertex AI to deploy a generative model for customer support. They need to ensure that the model does not produce harmful content when responding to user queries. Which of the following is the MOST effective approach?

Accepted Answer

Use Vertex AI's safety filters and adjust safety settings per category. Vertex AI's safety filters and adjustable safety settings are the built-in, most effective way to block harmful content. The other options are either insufficient or place undue burden on users.

Answer

Instruct customers to report harmful responses for future model updates

Answer

Prompt the model to 'be safe' and rely on its training to avoid harmful content

Answer

Review all responses manually before sending to customers

Question 7

An ML engineer sees the above deployment output. The business wants to reduce inference cost. Which action should they take?

Accepted Answer

Change to a lower-cost machine type. Switching to a lower-cost machine type directly reduces the per-request compute cost without altering the model architecture or inference logic. This is a common cost-optimization strategy in cloud-based ML deployments, where instance types (e.g., from GPU to CPU or from a larger to a smaller GPU) can be selected based on latency and throughput requirements, provided the model fits within the machine's memory and compute constraints.

Answer

Use a larger model

Answer

Deploy to multiple regions

Answer

Increase traffic split

Question 8

A manufacturing company wants to use generative AI to create maintenance manuals from sensor data. The manuals must be accurate and reflect the latest equipment configurations. Which approach best ensures data freshness and consistency?

Accepted Answer

Use a retrieval-augmented generation (RAG) system that queries a live database of sensor configurations.. A retrieval-augmented generation (RAG) system retrieves the most current equipment configurations directly from a live database at inference time, ensuring the generated manual reflects real-time sensor data without requiring model retraining. This approach decouples the static knowledge in the LLM from the dynamic data source, guaranteeing both accuracy and freshness while avoiding the latency and cost of continuous retraining.

Answer

Train the model in real-time as sensor data streams in.

Answer

Periodically retrain the model with the latest sensor data.

Answer

Have human technicians review and update the manuals manually.

Question 9

A financial services firm is using a foundation model on Vertex AI to generate investment summaries from quarterly reports. The summaries are accurate but often miss key financial metrics and trends. The team cannot afford to fine-tune the model frequently. Which technique should they use to improve the completeness and relevance of the summaries without modifying the model?

Accepted Answer

Provide three few-shot examples in the prompt that highlight the desired metrics.. Few-shot prompting provides the model with concrete examples of desired output structure and content, guiding it to include key financial metrics and trends without retraining. This technique leverages in-context learning, where the model generalizes from the examples in the prompt to produce more complete and relevant summaries, while avoiding the cost and latency of fine-tuning.

Answer

Increase temperature to 0.9 to encourage more creative outputs.

Answer

Set stop sequences to ['

'] to ensure the model finishes each paragraph.

Answer

Lower top_p to 0.5 to reduce the sampling pool.

Question 10

A developer is using the Vertex AI Gemini API to generate product descriptions. They get a 400 error 'INVALID_ARGUMENT: The model's maximum input token limit is 8192.' What is the most likely issue?

Accepted Answer

The prompt is too long. The 400 error 'INVALID_ARGUMENT: The model's maximum input token limit is 8192' explicitly indicates that the combined token count of the prompt (system instructions, user input, and any conversation history) exceeds the 8192-token context window of the Gemini model being used. This is a hard limit enforced by the Vertex AI Gemini API, and the error is triggered before any generation begins. Therefore, the most likely issue is that the prompt is too long.

Answer

The API key is invalid

Answer

The output tokens are too high

Answer

The model is not available in the region

Question 11

A company wants to build a customer support chatbot that answers based on internal documentation. They use Vertex AI Search and want to ensure the model only uses retrieved documents. What should they do?

Accepted Answer

Enable grounding with Vertex AI Search. Grounding with Vertex AI Search ensures the model's responses are strictly based on the retrieved documents from the internal documentation, preventing hallucination or reliance on pre-trained knowledge. Grounding works by providing the model with a search result context that it must use as the sole source for generating answers, effectively constraining the output to the provided documents.

Answer

Fine-tune the model on the documentation

Answer

Increase max output tokens

Answer

Set temperature to 0.0

Question 12

After fine-tuning a model on customer support data, the model starts using profanity. What is the most effective mitigation?

Accepted Answer

Enable a safety attribute filter. Enabling a safety attribute filter is the most effective mitigation because it acts as a post-processing guardrail that blocks profanity at inference time, regardless of the model's training data. This is a standard practice in production LLM deployments, where safety filters (e.g., using keyword matching or classifier models) intercept and redact harmful outputs before they reach the user, providing immediate and reliable control without requiring retraining.

Answer

Add profanity to training data as negative examples

Answer

Reduce learning rate and retrain

Answer

Increase temperature to reduce confidence

Question 13

A product manager wants to quickly build a conversational agent that can answer FAQs from the company's help center articles. They have limited coding experience. Which Google Cloud service is BEST suited for this task?

Accepted Answer

Vertex AI Agent Builder. Vertex AI Agent Builder is the best choice because it provides a no-code/low-code interface specifically designed for building conversational agents and search experiences. It allows the product manager to connect help center articles as a data source and automatically generate a FAQ-answering agent without writing code, making it ideal for someone with limited coding experience.

Answer

Vertex AI Pipelines

Answer

Vertex AI Studio

Answer

Model Garden

Question 14

Which Google Cloud AI service provides a unified ML platform for building, deploying, and managing ML models in production?

Accepted Answer

Vertex AI. Vertex AI is the correct answer because it is Google Cloud's unified ML platform that integrates data engineering, data science, and ML engineering workflows into a single service. It provides end-to-end capabilities for building, training, deploying, and managing ML models in production, including AutoML, custom training, model registry, and MLOps features like continuous evaluation and monitoring.

Answer

AI Platform

Answer

BigQuery ML

Answer

Cloud AutoML

Question 15

A marketing team wants to generate consistent brand-aligned social media posts using Vertex AI Studio. Which prompt engineering technique should they use to ensure the output tone matches their brand voice?

Accepted Answer

Provide a few-shot prompt with examples of previous brand-aligned posts. Providing a few-shot prompt with examples of previous brand-aligned posts (B) is the most effective technique because it shows the model concrete patterns of tone, style, and vocabulary, allowing it to generalize and produce consistent output. Few-shot prompting is ideal when the desired style is nuanced and hard to describe abstractly. This directly ensures the generated posts match the brand voice.

Answer

Set the temperature to 0 and use a long system instruction

Answer

Use a zero-shot prompt describing the brand voice

Answer

Use chain-of-thought prompting to explain the reasoning behind each post

Question 16

An enterprise deploys a large language model (LLM) for internal document summarization. Users complain that summaries sometimes include statements not present in the original document. Which mitigation strategy should the team prioritize to address this hallucination issue?

Accepted Answer

Implement retrieval-augmented generation (RAG) to ground the model in the original documents and require citations.. Retrieval-Augmented Generation (RAG) is the most direct and effective mitigation for hallucination in document summarization because it forces the LLM to base its output on retrieved chunks of the original document. By requiring citations, the model must reference specific passages, making it verifiable and reducing the likelihood of fabricating content. This grounds the generation in the source material, addressing the root cause of hallucination—lack of factual grounding—rather than relying on post-hoc correction or output tuning.

Answer

Train a discriminator model to detect hallucinations and perform adversarial training.

Answer

Apply reinforcement learning from human feedback (RLHF) using a reward model that penalizes hallucinations.

Answer

Reduce the model's temperature parameter to 0 to make outputs deterministic.

Question 17

A company is deploying a generative AI application using Vertex AI. They need to minimize latency for real‑time inference while maintaining high quality. Which TWO actions are most effective?

Accepted Answer

Use a smaller model like Gemini Flash. Option B is correct because Gemini Flash is a lower-latency, lighter-weight model in the Gemini family, purpose-built for high-throughput, real-time tasks, so it directly reduces inference latency while still delivering strong quality for most generative AI use cases. Option D is correct because output tokens are generated sequentially (autoregressive decoding), so capping max output tokens to the shortest acceptable length proportionally cuts generation time and thus end-to-end latency. Option A is not appropriate because batching requests improves throughput and cost efficiency, not per-request real-time latency, and can actually add queuing delay. Option C is wrong because Gemini Pro is larger and slower than Flash, trading latency for quality rather than minimizing latency. Option E is wrong because raising temperature to 1.0 only affects randomness/creativity of sampling and does not reduce latency.

Answer

Batch multiple inference requests together

Answer

Use Gemini Pro instead of Gemini Flash to ensure quality

Answer

Increase the temperature to 1.0 for more creative outputs

Question 18

A streaming platform uses a large generative model for personalized content suggestions. Budget constraints require minimizing inference costs without significantly degrading quality. Which approach is most effective?

Accepted Answer

Use a distilled version of the model.. Distillation trains a smaller 'student' model to mimic a larger 'teacher' model, reducing parameter count and inference latency while retaining most of the recommendation quality. This directly addresses the budget constraint by lowering compute and memory costs per inference, making it the most effective approach among the options.

Answer

Deploy the model on higher-end accelerators to save time.

Answer

Implement stronger safety filters to reduce output length.

Answer

Cache frequent prompts to avoid regeneration.

Question 19

A team is fine-tuning a large language model on custom data using Vertex AI. They find that the training loss decreases but validation loss increases. What is the best course of action?

Accepted Answer

Reduce the model size or add dropout regularization.. The increasing validation loss while training loss decreases is a classic sign of overfitting, where the model memorizes the training data but fails to generalize. Reducing model size or adding dropout regularization directly combats overfitting by limiting the model's capacity or introducing noise during training, which forces the model to learn more robust features. This is the best course of action because it addresses the root cause without further exacerbating the problem.

Answer

Increase the number of training epochs.

Answer

Increase the learning rate.

Answer

Switch to a smaller batch size.

Question 20

A machine learning engineer wants to reduce the latency of a Gemini-based chatbot running in production. Which TWO strategies would be MOST effective?

Accepted Answer

Switch from Gemini Pro to Gemini Flash. Option A is correct because Gemini Flash is a lighter, lower-latency model variant than Gemini Pro, so switching to it directly reduces the time the model takes to generate each response. Option B is correct because max_output_tokens caps the number of tokens the model generates, and since generation time scales roughly with output length, lowering this value shortens the response and thus the overall latency. Option C is not correct because streaming improves perceived latency by delivering tokens incrementally, but it does not reduce the actual total generation time. Option D is not correct because fine-tuning adapts the model to a task and does not inherently lower inference latency. Option E is not correct because raising temperature to 1.0 affects randomness and creativity, not speed, and can even make outputs longer or less predictable.

Answer

Enable streaming mode

Answer

Fine-tune the model on the specific task

Answer

Increase the temperature to 1.0