GENERATIVE-AI-LEADER Practice Questions

The core requirement is to summarize breaking news from various online sources in near real-time. Grounding with Google Search is a feature in Vertex AI that connects a large language model (LLM) to the live Google Search index. This allows the model to access and process the most current information available on the web. By grounding the summarization model in Google Search, the news agency can ensure its generated content is timely, factually based on the latest reports, and less prone to hallucinations, which is critical for reporting on fast-developing stories.

A. Document AI is designed for extracting structured data from unstructured or semi-structured documents like invoices and forms, not for summarizing real-time web content.

B. BigQuery is a serverless data warehouse for large-scale data analytics and querying stored datasets, not a generative AI tool for text summarization.

C. Vertex AI Natural Language API offers pre-trained models for tasks like sentiment analysis and entity extraction but does not inherently provide the real-time web-crawling and indexing capability required for breaking news.

1. Google Cloud Documentation

"Grounding in Vertex AI": "Grounding is the ability to connect model-generated content to verifiable sources of information... For models that support grounding with Google Search

the model can access information from Google Search to generate a response. The grounded response includes citations to the sources that the model used to generate its response." This directly supports using Google Search to access current

verifiable information for generation tasks.

2. Google Cloud Documentation

"Overview of Vertex AI Search": "You can ground your results on public data by using Google Search... This lets you build applications that can answer questions about a wide range of topics by using information from the public web." This highlights its suitability for applications needing access to broad

up-to-date public information

like a news agency.

3. Google Cloud Blog

"How grounding improves LLM performance for complex

knowledge-intensive tasks" (May 23

2023): The article explains

"Grounding helps to ensure that the model's output is relevant

accurate

and helpful... For example

a news organization could use a grounded LLM to generate summaries of breaking news stories." This provides a direct use-case example matching the question's scenario.

Security Command Center is Google Cloud's centralized security and risk management platform. It provides a comprehensive, single-pane-of-glass view of security health across an entire Google Cloud organization. It aggregates security findings from various built-in and third-party services, enabling teams to discover assets, detect vulnerabilities, and identify threats, including those that might affect generative AI deployments. This centralization is precisely what the security team in the scenario requires for a holistic overview and effective risk management.

A. Workload monitoring tools, like Cloud Monitoring, focus on operational performance, uptime, and resource utilization, not on centralized security threat and vulnerability management.

C. Identity and Access Management (IAM) is a foundational service for controlling who can do what on specific resources, not a platform for a comprehensive security overview.

D. Secure-by-design infrastructure refers to the inherent security built into Google's global infrastructure, not a tool or platform that a security team uses for monitoring.

1. Google Cloud Documentation - Security Command Center overview: "Security Command Center is a centralized vulnerability and threat reporting service that helps you strengthen your security posture by identifying misconfigurations

vulnerabilities

and threats across your Google Cloud organization." (Source: Google Cloud

Security Command Center Documentation

"Overview of Security Command Center"

Section: "How Security Command Center works").

2. Google Cloud Whitepaper - Google Security Whitepaper: "Security Command Center provides a single pane of glass for security visibility

compliance

and governance across your Google Cloud services." (Source: Google Cloud

"Google Security Whitepaper"

2020

Page 35

Section: "Visibility and control").

3. Google Cloud Documentation - IAM overview: "Cloud Identity and Access Management (IAM) lets you grant granular access to specific Google Cloud resources and helps prevent access to other resources." This highlights its focus on access control

not centralized monitoring. (Source: Google Cloud

IAM Documentation

"Overview of IAM").

4. Google Cloud Documentation - Cloud Monitoring overview: "Cloud Monitoring collects metrics

events

and metadata from Google Cloud services

hosted uptime probes

application instrumentation

and a variety of common application components." This confirms its focus on operational metrics. (Source: Google Cloud

Cloud Monitoring Documentation

"Cloud Monitoring overview").

The most direct way for a development team with limited AI experience to launch a solution quickly is by leveraging pre-built components and simplified development environments. Google Cloud's pre-trained foundation models (like Gemini) provide a powerful starting point, eliminating the need for model creation from scratch. Furthermore, low- and no-code tools, such as Vertex AI Studio for prompt design and Vertex AI Search for building search and conversational agents, abstract away the underlying complexity. This allows developers to focus on application logic and user experience, significantly accelerating the development-to-launch timeline without requiring deep AI/ML expertise.

B. Community and forums are valuable for support and troubleshooting but do not provide the core tools needed to build and launch an application quickly.

C. Training materials are essential for upskilling but require a time investment, which conflicts with the immediate goal of launching the solution quickly.

D. Access to the latest tools is a general benefit, but it doesn't specifically address the team's lack of experience or the need for rapid development, which low-code tools and pre-trained models do.

1. Google Cloud Documentation

"Generative AI on Vertex AI": "Build with foundation models. Access Google’s state-of-the-art foundation models

including Gemini

Imagen

and more

on Vertex AI. With the Vertex AI platform

you can tune

deploy

and govern models to build and deploy gen AI applications quickly and responsibly." This highlights how pre-trained (foundation) models enable quick development.

Source: Google Cloud

Generative AI on Vertex AI

"Build with foundation models" section.

2. Google Cloud Documentation

"Vertex AI Studio": "Vertex AI Studio is a Google Cloud console UI for rapidly prototyping and testing generative AI models." The emphasis on a UI for "rapidly prototyping" supports the idea that these tools are designed for speed and ease of use

lowering the barrier for developers.

Source: Google Cloud

Overview of Vertex AI Studio

"Introduction" paragraph.

3. Google Cloud Documentation

"Vertex AI Search": "Vertex AI Search lets developers with little ML expertise quickly build Google-quality search experiences in their websites and applications." This explicitly states that the tool is for developers with limited expertise and for quick implementation.

Source: Google Cloud

Vertex AI Search documentation

"Overview" section.

4. Google Cloud Skills Boost

"Introduction to Generative AI" Course: The courseware frequently emphasizes how Google's managed services and foundation models democratize AI. For example

the module "Generative AI on Google Cloud" explains that Vertex AI provides "a unified platform to build

deploy

and manage ML models

" which includes "access to foundation models" and tools that "accelerate the development of AI applications."

Source: Google Cloud Skills Boost

Introduction to Generative AI

Module: "Generative AI on Google Cloud".

An AI agent, in the context of generative AI, is an autonomous system that leverages a large language model (LLM) as its core reasoning engine. Its primary function is to perceive a situation or user request, break it down into actionable steps, and then select and utilize a variety of "tools" (such as APIs, search engines, or other models) to accomplish a complex goal. This process involves analysis, planning, and decision-making to act on behalf of a user with minimal human intervention, distinguishing it from a simple conversational interface or underlying infrastructure.

A. This describes the function of cloud infrastructure components like GPUs and TPUs, which an agent uses but does not provide.

B. This describes the function of data storage and management services (e.g., Google Cloud Storage, BigQuery), not the agent's role.

C. This describes a user interface (UI). While an agent may have a UI, its core function is the autonomous reasoning and task execution that happens behind the interface.

1. Google Cloud Blog

"What are AI agents?" (May 2024): This article defines an AI agent as "a software program that can perform tasks on behalf of a user with some level of autonomy." It explicitly states that agents use LLMs for reasoning and can "access a set of tools to perform actions in the world

" which directly supports the concept of analyzing situations and using multiple tools to make decisions.

2. Google Cloud Documentation

"Overview of Agent Builder": The documentation for Vertex AI Agent Builder describes how it enables the creation of agents that "can hold conversations" and are "equipped with tools" that allow them to "interact with external systems" to fulfill complex user requests. This highlights the agent's function in orchestrating tools to perform actions

aligning with the correct answer.

3. Stanford University

CS221: Artificial Intelligence: Principles and Techniques

Autumn 2021

Lecture 1: Course materials define an agent as an entity that "perceives its environment through sensors and acts upon that environment through actuators." In the context of Gen AI

the LLM acts as the reasoning component

and "tools" are the actuators

allowing the agent to make informed decisions and act autonomously. This foundational definition supports the core concepts in option D.

Prompt chaining is the technique of using the output of one prompt as the input for a subsequent prompt. In the context of a conversational chatbot, this is essential for maintaining state and memory. To ensure the chatbot can access previous portions of the conversation, the application must chain the prompts by appending the new user query to the existing conversation history. This combined text is then sent to the model as the new prompt, allowing it to generate a response that is contextually aware of the entire interaction.

A. Use zero-shot prompting.

This technique asks the model to perform a task without any prior examples in the prompt. It does not inherently manage or access conversational history across multiple turns.

B. Use role prompting.

This sets the persona or character of the AI (e.g., "You are a helpful bank teller"). While useful for tone, it is not the mechanism for remembering past conversational context.

C. Use few-shot prompting.

This provides a few examples of the desired input/output format within a single prompt to guide the model. It does not maintain the history of a multi-turn conversation.

---

1. Google Cloud Documentation

"Introduction to prompt design": This document defines prompt chaining: "Chaining is the practice of using the output of one prompt as the input of a second prompt. You can create sophisticated applications by chaining prompts together." This principle is directly applied in multi-turn conversations to maintain context.

2. Google Cloud Skills Boost

"Introduction to Large Language Models": In the module on "Prompt Engineering

" the course explains that for multi-turn conversations

the prompt must be updated with the conversation history. It states

"To have a multi-turn conversation

you need to send the conversation history with each new turn." This is a direct application of the prompt chaining concept for building chatbots.

3. Google Cloud Blog

"An introduction to prompt design for generative AI": This article discusses building conversational experiences and notes the importance of including conversation history in the prompt. "For a chatbot to have a memory of the conversation

you need to include the conversation history in the prompt." This confirms that managing history is a key prompting strategy

achieved through chaining.

Google Cloud's Identity and Access Management (IAM) is the primary service for implementing fine-grained access control. It allows organizations to manage authorization by defining "who" (identity) has "what access" (role) for "which resource." For generative AI resources, such as models in Vertex AI, IAM enables administrators to grant specific permissions like viewing, using, or managing models to specific users, groups, or service accounts, thereby providing the required granular control.

B. Secure-by-design infrastructure: This is a foundational principle of how Google Cloud is built, encompassing physical security and data encryption, not a specific tool for managing user access to AI models.

C. Security Command Center: This is a centralized security management and risk platform for threat detection and posture assessment, not for setting granular user permissions on resources.

D. Workload monitoring tools: These tools, like Cloud Monitoring, are for observing performance and operational health. They do not manage or enforce access control policies.

1. Google Cloud Documentation

"Overview of IAM": "Cloud Identity and Access Management (IAM) lets you grant granular access to specific Google Cloud resources and helps prevent access to other resources. IAM lets you adopt the security principle of least privilege

so you grant only the necessary access to your resources." (cloud.google.com/iam/docs/overview)

2. Google Cloud Documentation

"Vertex AI access control with IAM": "You can control access to Vertex AI resources by using Identity and Access Management (IAM). IAM lets you grant access to specific Google Cloud resources and prevents unwanted access to other resources... For example

you can grant a user the AI Platform Viewer (roles/aiplatform.viewer) role so the user can view Vertex AI resources in a project." (cloud.google.com/vertex-ai/docs/general/access-control)

3. Google Cloud Security Whitepaper

"Google Cloud security foundations guide": This guide describes the "secure-by-design" approach as a layered security strategy built into the infrastructure

from the physical data centers to the purpose-built security hardware. This confirms it is a design philosophy

not an access control service. (cloud.google.com/resources/google-cloud-security-foundations-guide

Section: "Secure-by-design infrastructure")

The marketing team requires a versatile solution to generate both written articles (text) and images from text. Gemini is Google's family of natively multimodal models, designed from the ground up to understand, operate across, and combine different types of information, including text and images. This inherent multimodality makes it the ideal choice for a team without deep AI expertise that needs to create diverse campaign assets like articles and ad visuals using a single, powerful model family.

A. Gemma models are open, lightweight models primarily for text generation, aimed at developers, and do not natively generate images.

B. Imagen is a specialized text-to-image model and cannot be used to generate written articles, failing the versatility requirement.

D. Veo is a text-to-video generation model and is not designed for creating written articles or static images.

1. Gemini: Google AI. (2023

December 6). Introducing Gemini: our largest and most capable AI model. "Gemini is our first model to be multimodal from the ground up... This makes Gemini natively multimodal and able to seamlessly understand

operate across and combine different types of information including text

code

audio

image and video." This source confirms Gemini's native ability to handle both text and image modalities

fulfilling the core requirement.

(Source: Google AI Blog

"Introducing Gemini")

2. Imagen: Google Cloud. Image generation with Imagen on Vertex AI. "Imagen on Vertex AI lets you generate and edit studio-grade images at scale." This official documentation specifies Imagen's purpose is image generation

not text creation.

(Source: Google Cloud Documentation

Vertex AI Model Garden)

3. Gemma: Google AI. (2024

February 21). Gemma: Introducing new state-of-the-art open models. "Gemma is a family of lightweight

state-of-the-art open models... available in two sizes: Gemma 2B and Gemma 7B. Each size is released with pre-trained and instruction-tuned variants." The documentation highlights Gemma as a family of text-based models for developers.

(Source: Google AI Blog

"Gemma: Introducing new state-of-the-art open models")

4. Veo: Google DeepMind. (2024

May 14). Veo: our most capable video generation model. "Today

we’re announcing Veo

our most capable video generation model... Veo generates high-quality

1080p resolution videos that can go beyond a minute..." This source clearly defines Veo's function as video generation

which is not what the marketing team requires.

(Source: Google DeepMind Blog

"Veo: our most capable video generation model")

The tasks described—summarizing security case information, researching threats, and creating detection rules—are core functions of cybersecurity operations. A "Security agent" is a specialized generative AI tool designed specifically for these security-centric use cases. It integrates with security platforms like Security Information and Event Management (SIEM) and Security Orchestration, Automation, and Response (SOAR) to automate and augment the work of security analysts, directly addressing the CISO's requirements.

B. Data agent: This is too general. It focuses on broad data analysis and querying across various domains, not the specialized field of threat intelligence and response.

C. Code agent: This is specialized for software development tasks like writing, debugging, and explaining code, not for security-specific rule creation or threat research.

D. Customer service agent: This is designed for customer interactions and support, which is entirely unrelated to the internal security operations described in the scenario.

1. Google Cloud. (n.d.). Chronicle Security Operations. Retrieved from https://cloud.google.com/chronicle. The documentation for Chronicle CyberShield

a core component of the platform

details its generative AI capabilities: "Chronicle’s Gen AI capabilities can help you with natural language search

case summarization

and remediation recommendations

" which are the exact functions of a security agent as described in the question.

2. Google Cloud Blog. (2023

August 29). Introducing the Google Cloud Security AI Workbench. Retrieved from https://cloud.google.com/blog/products/security/introducing-google-cloud-security-ai-workbench. This official announcement describes the workbench's purpose

which embodies a security agent: "Security AI Workbench uses a specialized

security-trained large language model (LLM)

Sec-PaLM 2... to help with threat research

summarizing case information

and recommending actions."

3. Google Cloud. (n.d.). Security AI. Retrieved from https://cloud.google.com/security/ai. This page outlines Google's vision for AI in security

stating it helps "automatically generate detections... summarize event data... and recommend responses." These are the defining characteristics of a security-focused AI agent.

The scenario describes a classic application of reinforcement learning (RL). In RL, an agent (the AI character) learns to make decisions by performing actions in an environment (the video game) to maximize a cumulative reward. The system of positive and negative scores directly corresponds to the reward/penalty mechanism central to RL. The agent learns an optimal policy through trial and error, associating actions with long-term rewards, which perfectly aligns with the described process of improving the character's ability to play the game effectively.

B. Unsupervised learning: This approach is used to find hidden patterns or intrinsic structures in unlabeled data, such as clustering similar items. It does not involve an agent learning from rewards or penalties.

C. Supervised learning: This requires a pre-existing, labeled dataset where correct outputs are known for given inputs. The AI character is learning from experience, not from a dataset of "correct" moves.

D. Deep learning: This is a set of techniques using neural networks, not a fundamental learning paradigm like RL. While deep learning can be used to implement an RL agent (Deep Reinforcement Learning), RL is the overarching methodology described.

1. Google Cloud Documentation: "Introduction to reinforcement learning" on Vertex AI states

"Reinforcement learning (RL) is a machine learning technique that trains a software agent to make decisions by performing actions in an environment to maximize a cumulative reward... RL is a good choice for problems that require making a sequence of decisions to achieve a goal

such as playing a game..."

Source: Google Cloud. (n.d.). Introduction to reinforcement learning. Vertex AI Documentation. Retrieved from https://cloud.google.com/vertex-ai/docs/start/introduction-reinforcement-learning. (Refer to the "Overview" section).

2. Academic Publication (Foundational Text): Sutton

R. S.

& Barto

A. G. (2018). Reinforcement Learning: An Introduction. MIT Press.

Details: Chapter 1

Section 1.1

"Reinforcement Learning

" describes the core concept: "...how an agent ought to take actions in an environment so as to maximize some notion of cumulative reward. Reinforcement learning is one of three basic machine learning paradigms

alongside supervised learning and unsupervised learning." This directly maps to the game character scenario.

3. University Courseware: Stanford University

CS 234: Reinforcement Learning.

Details: The course overview defines the RL problem as an agent interacting with an environment over time

choosing actions to maximize a scalar reward signal. This is precisely the problem faced by the AI character in the video game. (Refer to Winter 2022

Lecture 1 slides

"What is Reinforcement Learning?").

The primary function of safety settings in a generative AI model is to implement responsible AI practices by filtering potentially harmful content. These settings allow developers to adjust the thresholds for various safety categories, such as harassment, hate speech, sexually explicit material, and dangerous content. By configuring these filters, the model can block responses that are likely to be inappropriate or unsafe, which is a critical requirement for customer-facing applications. This ensures that the generated output aligns with safety policies and protects end-users from harmful interactions.

A. Limiting text length is controlled by parameters like maxoutputtokens, not safety settings.

B. The number of tokens a model can process relates to its context window, a fundamental model attribute, not a configurable safety filter.

D. Controlling creativity and randomness is the function of parameters such as temperature, top-k, and top-p.

1. Google Cloud Documentation

"Safety filter thresholds": "Use safety filters to adjust the likelihood of getting responses that may be harmful. By default

safety filters block content...with a medium or high probability of being unsafe across all safety dimensions." This document details the categories (Harassment

Hate speech

etc.) and block thresholds

directly supporting the function described in option C.

Source: Google Cloud

Vertex AI Documentation

"Gemini API"

"Safety settings".

2. Google Cloud Documentation

"Model parameters": This documentation explicitly separates parameters for controlling creativity (temperature

topK

topP) and response length (maxOutputTokens) from safety settings

confirming they serve different primary functions.

Source: Google Cloud

Vertex AI Documentation

"Generative AI on Vertex AI"

"Model parameters".

3. Google AI

"A new generation of models for safety": "We’ve also built dedicated safety classifiers to identify

label

and stop harmful content... These classifiers are part of a multi-layered safety strategy." This highlights that safety is a distinct

classified filtering function.

Source: Google AI

Responsible AI

"A new generation of models for safety" section.