Generative AI Leader Practice Test 18 — 15 Questions

Question 1

A project manager wants to understand which Google Cloud generative AI services are subject to the 'Prohibited Use' policy. Where can they find the most up-to-date information?

Accepted Answer

The Google Cloud Acceptable Use Policy. The Google Cloud Acceptable Use Policy (AUP) is the authoritative document that defines prohibited uses of Google Cloud services, including generative AI offerings. It is regularly updated to reflect current legal, ethical, and security requirements, making it the most reliable source for the most up-to-date information on prohibited use cases. The AUP explicitly covers restrictions on generating harmful content, engaging in illegal activities, and violating intellectual property rights, which directly apply to generative AI services.

Answer

Google Cloud documentation

Answer

Google's AI Principles

Answer

The Gemini Terms of Service

Question 2

A developer is building a real-time speech transcription application for customer support calls. The audio is streamed, and the transcription must be returned with low latency. Which Google Cloud AI service should they use?

Accepted Answer

Cloud Speech-to-Text with streaming recognition. Cloud Speech-to-Text supports streaming recognition via gRPC or the streaming REST endpoint, returning partial transcripts as audio arrives, which is exactly what low-latency real-time transcription requires. It handles telephony audio with models tuned for phone calls and supports interim results for responsiveness.

Answer

Natural Language API

Answer

Cloud Text-to-Speech

Answer

Vertex AI with a custom model

Question 3

The exhibit shows the output of describing a model on Vertex AI. What does 'modelSource: MODEL_GARDEN' indicate about this model?

Accepted Answer

The model was imported from the Vertex AI Model Garden.. 'modelSource: MODEL_GARDEN' explicitly indicates that the model was sourced from Vertex AI Model Garden, which is a curated repository of pre-built and pre-trained foundation models. This field is set when a model is imported from Model Garden, not when it is trained or fine-tuned from scratch within Vertex AI.

Answer

The model was trained on Vertex AI from scratch.

Answer

The model has been exported to Model Garden.

Answer

The model was fine-tuned using AutoML.

Question 4

An organization needs to deploy a generative AI application with strict compliance requirements, including data residency and auditability of model decisions. Which Google Cloud feature should they prioritize?

Accepted Answer

Vertex AI. Vertex AI is Google Cloud's enterprise ML platform that provides data residency controls (regional endpoints, VPC Service Controls), auditability via Cloud Audit Logs for every prediction, model governance features, and integration with Model Armor for safety. It is the only option that natively supports the compliance, residency, and audit requirements described.

Answer

Colab Enterprise

Answer

Gemini API

Answer

Model Garden

Question 5

A developer wants to use a pre-trained model to identify objects in images. Which Google Cloud AI API should they use?

Accepted Answer

Vision AI. Vision AI provides pre-trained models for object detection, image classification, etc. Natural Language AI is for text, Speech-to-Text for audio, and Translation for text translation.

Answer

Speech-to-Text

Answer

Translation API

Answer

Natural Language AI

Question 6

A company is building a generative AI chatbot for customer support using Vertex AI. They want to ground the model responses with their internal knowledge base stored in Cloud Storage and BigQuery. Which feature should they use to ensure the model only answers from the provided data and avoids hallucination?

Accepted Answer

Vertex AI Grounding with Vertex AI Search. Vertex AI Grounding with Vertex AI Search is the correct feature because it allows the model to retrieve and cite information from a specified data source (such as Cloud Storage and BigQuery) to generate responses. This process, known as grounding, ensures the model's output is based solely on the provided authoritative data, effectively reducing hallucinations by constraining the model to factual, retrieved content rather than relying on its internal parametric knowledge.

Answer

Vertex AI Prediction

Answer

Vertex AI Pipelines

Answer

Cloud Functions

Question 7

A developer is using the Gemini API to classify customer emails. They want to ensure that the model always returns one of three predefined labels: 'complaint', 'inquiry', or 'feedback'. Which model configuration is MOST appropriate?

Accepted Answer

Set temperature to 0.0 and use few-shot examples with required labels in the prompt. Setting temperature to 0.0 makes the model deterministic, minimizing randomness and ensuring consistent output. Combined with few-shot examples that explicitly list the three required labels ('complaint', 'inquiry', 'feedback') in the prompt, this configuration reliably constrains the model to return only those labels, which is the most appropriate approach for a strict classification task.

Answer

Set temperature to 1.0 and top-p to 0.9 to allow creativity while constraining via system instructions

Answer

Fine-tune the model on a dataset of labeled emails to memorize the three classes

Answer

Use top-k sampling with k=50 and no temperature adjustment

Question 8

An organization is building a RAG system using Vertex AI Vector Search. They notice that the retrieved documents are not relevant to the user's query. What is the most likely cause?

Accepted Answer

The embedding model used does not capture the semantic meaning of the documents effectively. The most likely cause is that the embedding model fails to map the semantic meaning of the documents and queries into a shared vector space effectively. In Vertex AI Vector Search, retrieval quality depends entirely on the cosine similarity between query and document embeddings; if the embeddings are poor, even a perfect vector index will return irrelevant results.

Answer

The context window of the LLM is too small

Answer

The chunk size of the documents is too large

Answer

The temperature setting in the LLM is too high

Question 9

A startup wants to embed generative AI features into their mobile app but has limited ML expertise. Which Google Cloud service is best suited for rapid integration with no ML training?

Accepted Answer

Gemini API. The Gemini API provides direct, no-code access to Google's most capable generative AI models via a simple REST API, requiring zero ML training or infrastructure setup. This makes it the fastest path for a startup with limited ML expertise to embed generative AI features like text generation, summarization, or chat into a mobile app.

Answer

Vertex AI Model Garden

Answer

Vertex AI Agent Builder

Answer

Cloud Run with a custom container

Question 10

A company is transitioning a generative AI pilot to production. They need to ensure cost predictability and scalability. Which THREE actions should they take?

Accepted Answer

Implement response caching for common queries. Response caching for common queries reduces latency and API costs by serving repeated requests from a cache instead of invoking the generative AI model each time. This directly improves cost predictability (fewer model invocations) and scalability (reduced load on the model endpoint), making it a core optimization for production deployments.

Answer

Provision reserved throughput for all requests

Answer

Conduct A/B testing on model versions

Question 11

A healthcare startup is developing a generative AI system to assist doctors in diagnosing rare diseases. According to Google's AI Principles, what is the MOST important requirement before deployment?

Accepted Answer

The system must include a mechanism for human review of all diagnostic suggestions. Google's AI Principles emphasize that AI systems should be socially beneficial and avoid creating or reinforcing unfair bias, and for high-stakes domains like healthcare, human oversight is critical. The principle of 'be accountable to people' requires that AI systems include mechanisms for human review and feedback, especially for diagnostic suggestions that could affect patient safety.

Answer

The model must achieve at least 99% accuracy on a held-out test set

Answer

The model must be trained on the most recent medical literature

Answer

The startup must publish the model's architecture in a peer-reviewed journal

Question 12

A researcher wants to use Google's AlphaFold for a project. What is the primary capability of AlphaFold?

Accepted Answer

Predicting 3D protein structures from amino acid sequences. AlphaFold, developed by Google DeepMind, is specifically designed to predict the 3D structure of proteins from their amino acid sequences. This capability solves a fundamental challenge in biology, as the function of a protein is largely determined by its 3D shape, and experimental methods like X-ray crystallography are time-consuming and expensive. AlphaFold achieves this using a deep learning architecture that integrates multiple sequence alignment (MSA) and pairwise distance predictions to model the spatial coordinates of atoms.

Answer

Generating realistic human speech

Answer

Playing the game of Go at superhuman level

Answer

Generating code from natural language descriptions

Question 13

What is the primary benefit of using embeddings and vector search in a generative AI application?

Accepted Answer

They enable efficient retrieval of semantically similar content. Embeddings convert text into dense vector representations that capture semantic meaning, and vector search enables efficient retrieval of semantically similar content by finding nearest neighbors in vector space. This retrieval-augmented generation (RAG) approach grounds the generative AI model in relevant external knowledge, improving accuracy and reducing hallucinations without retraining.

Answer

They improve the model's ability to generate code

Answer

They reduce the size of the model by compressing weights

Answer

They allow the model to process images directly

Question 14

A data scientist is fine-tuning a large language model for a specialized domain using limited labeled data. To avoid catastrophic forgetting and reduce computational cost, which approach is recommended?

Accepted Answer

Adapter-based fine-tuning using LoRA. Adapter-based fine-tuning using LoRA (Low-Rank Adaptation) is recommended because it freezes the pre-trained model weights and injects trainable low-rank matrices into the transformer layers. This approach drastically reduces the number of parameters to update (often by 10,000x), lowering memory and compute requirements, while preserving the original knowledge to prevent catastrophic forgetting on limited domain data.

Answer

Using prompt engineering with in-context learning

Answer

Full fine-tuning of all model parameters

Answer

Training a new model from scratch on the domain data

Question 15

A financial services firm is using Vertex AI to generate investment reports. They need to ensure that the model outputs are explainable and comply with regulatory requirements. Which Vertex AI feature should they use?

Accepted Answer

Vertex Explainable AI. Vertex Explainable AI provides feature attributions and explanations for model predictions, which is essential for financial services firms that must comply with regulatory requirements like the EU's GDPR or the US SEC's model risk management guidelines. It helps auditors and stakeholders understand why a model generated a specific investment report output, ensuring transparency and accountability in AI-driven decisions.

Answer

Vertex AI Model Registry

Answer

Vertex AI Safety Settings

Answer

Vertex AI AutoML