Generative AI Leader Practice Test 44 — 15 Questions

Question 1

A data analyst needs to run a simple regression model directly on data stored in BigQuery without moving data to another platform. Which service should they use?

Accepted Answer

BigQuery ML. BigQuery ML (B) is correct because it allows users to create and execute machine learning models using standard SQL syntax directly on data stored in BigQuery, without needing to export data to a separate platform. This service is specifically designed for running regression, classification, and other models natively within BigQuery, leveraging its serverless architecture and built-in ML capabilities.

Answer

TensorFlow on Compute Engine

Answer

Vertex AI Training

Answer

Google Colab

Question 2

Which TWO are components of the Vertex AI Generative AI Studio?

Accepted Answer

Model Garden. Model Garden (B) is a core component of Vertex AI Generative AI Studio, providing a curated catalog of foundation models from Google and third parties (e.g., PaLM, Gemini, Llama) that users can browse, test, and deploy. Prompt Editor (E) is also a built-in Studio component that lets users interactively design, test, and refine prompts against foundation models without writing code. Dataflow (A) is a managed Apache Beam service for batch and streaming data pipelines, unrelated to the Studio's model/prompt tooling. Pipeline templates (C) belong to Vertex AI Pipelines for orchestrating ML workflows, not the Generative AI Studio interface. Cloud Functions (D) is a serverless compute service for event-driven code, not a Generative AI Studio component.

Answer

Dataflow

Answer

Pipeline templates

Answer

Cloud Functions

Question 3

A company wants to use GenAI to automate customer support. They have a large knowledge base. Which approach maximizes ROI in the first 6 months?

Accepted Answer

Use a pre-built conversational AI platform with Retrieval-Augmented Generation (RAG). Maximizes ROI in the first 6 months because it leverages a pre-built conversational AI platform integrated with Retrieval-Augmented Generation (RAG). RAG allows the model to dynamically retrieve relevant information from the existing knowledge base at inference time, providing accurate, context-aware responses without the need for costly retraining or custom model development. This approach balances rapid deployment, low upfront investment, and high accuracy, making it the most cost-effective solution for automating customer support quickly.

Answer

Deploy a general-purpose chatbot without customization

Answer

Build a custom LLM from scratch using their data

Answer

Fine-tune a foundation model on historical support tickets

Question 4

Which THREE of the following are common techniques to reduce harmful biases in generative AI models? (Choose three.)

Accepted Answer

Use reinforcement learning from human feedback (RLHF) with a reward model that penalizes biased or unfair outputs.. Option A is correct because RLHF with a reward model that explicitly penalizes biased or unfair outputs directly optimizes the generative model's policy toward fairness-aligned behavior, using human preference data to shape the reward signal. Option B is correct because curating diverse and balanced training datasets—deliberately overrepresenting underrepresented groups—counteracts sampling bias and spurious correlations learned from skewed data, which is a foundational bias-mitigation technique. Option D is correct because adversarial training that removes protected attribute information from hidden representations (e.g., via adversarial classifiers or fair representation learning) reduces the model's ability to encode and exploit sensitive attributes. Option C does not belong because lowering the temperature only makes sampling more deterministic; it changes randomness, not the underlying learned biases, and can even amplify the most probable (biased) output. Option E does not belong because a legal review of outputs is a post-hoc compliance check, not a technique that reduces bias within the generative model itself.

Answer

Decrease the model's temperature parameter to make outputs more deterministic.

Answer

Conduct a legal review of all generated outputs before release.

Question 5

A company wants to ground Gemini responses with real-time Google Search results to improve accuracy of current events. Which feature enables this?

Accepted Answer

Google Search grounding. Google Search grounding connects Gemini to live search results to reduce hallucinations and provide up-to-date information.

Answer

BigQuery ML

Answer

Vertex AI Vector Search

Answer

Gemini API function calling

Question 6

A company is designing a prompt engineering strategy for a customer service chatbot using Gemini. Which two practices are recommended for improving response quality? (Choose TWO)

Accepted Answer

Use chain-of-thought prompting. Option A (Use chain-of-thought prompting) is correct because prompting Gemini to reason step by step before producing a final answer improves accuracy and coherence on complex customer service queries, especially those requiring multi-step logic or policy lookups. Option E (Include a system instruction to define the role) is correct because a system instruction sets persistent behavioral context, such as "You are a helpful customer service agent for X company," which anchors tone, scope, and constraints across all turns without repeating them in every user prompt. Option B is not recommended as a blanket rule: few-shot examples can help, but "always" providing multiple examples wastes tokens, increases latency, and can bias or overfit responses when zero-shot or single-example prompting suffices. Option C is wrong because omitting context strips the model of the grounding information needed to answer domain-specific questions accurately. Option D is wrong because temperature 1.0 increases randomness and creativity, which is undesirable for a customer service chatbot where factual, consistent, and deterministic answers are required; lower temperatures (e.g., 0.2-0.4) are typically preferred.

Answer

Always provide multiple examples in the prompt

Answer

Avoid any context in the prompt

Answer

Set temperature to 1.0 for maximum creativity

Question 7

Which TWO are benefits of using retrieval-augmented generation (RAG) over fine-tuning?

Accepted Answer

No need for training. Option A is correct because RAG does not modify the model's weights; it retrieves relevant documents at inference time and injects them into the prompt, so no training or fine-tuning run is required to add new knowledge. Option C is correct because RAG can pull from an external, continuously updated knowledge base or index, letting the system answer with information newer than the model's static training cutoff, whereas fine-tuning bakes in knowledge only as of the training data used. Options B, D, and E are not benefits guaranteed by RAG: it does not yield higher accuracy on all tasks (retrieval quality can introduce errors), it does not reduce the underlying model's size since the same LLM is still used, and it typically adds retrieval overhead that can increase, not lower, latency.

Answer

Higher accuracy on all tasks

Answer

Reduced model size

Answer

Lower latency

Question 8

Which of the following is a key principle in Google's AI Principles that directly addresses the need to avoid creating or reinforcing unfair bias?

Accepted Answer

Avoid creating or reinforcing unfair bias. Google's AI Principles explicitly state 'Avoid creating or reinforcing unfair bias' as a standalone principle. This principle directly mandates that AI systems must be designed and tested to mitigate biases in training data, model outputs, and deployment contexts, ensuring fairness across demographic groups. It is the most direct response to the question's focus on avoiding unfair bias.

Answer

Uphold high standards of scientific excellence

Answer

Be socially beneficial

Answer

Be accountable to people

Question 9

A company is deciding between building a custom fine-tuned model vs. using a pre-built API for a document summarization task. The documents contain domain-specific jargon. Which factor STRONGLY favors using a pre-built API with prompt engineering?

Accepted Answer

The requirement for low initial development cost and fast time-to-market. Using a pre-built API with prompt engineering eliminates the need for expensive model training infrastructure and specialized ML expertise, enabling rapid deployment at low initial cost. For a document summarization task, prompt engineering can leverage the API's existing capabilities without custom fine-tuning, making it ideal when speed and budget are primary constraints.

Answer

The need to handle highly specialized industry terminology

Answer

The need for the model to learn a unique writing style from past summaries

Answer

The need to keep all data on-premises for security compliance

Question 10

A company is running a GenAI proof-of-concept (PoC) for internal document Q&A. The PoC shows high latency and cost. The team suspects they are using an unnecessarily large model for the task. What is the BEST action to optimize?

Accepted Answer

Switch to a smaller model from Vertex AI Model Garden. Switching to a smaller model from Vertex AI Model Garden directly addresses the root cause of high latency and cost: an unnecessarily large model. Smaller models have fewer parameters, requiring less compute per inference, which reduces both response time and operational expense while often being sufficient for domain-specific tasks like internal document Q&A.

Answer

Disable grounding and rely solely on the model's internal knowledge

Answer

Increase the batch size for requests

Answer

Fine-tune the existing model on the company's documents

Question 11

A developer is using Vertex AI PaLM API to generate code snippets. The responses sometimes contain security vulnerabilities. What is the best practice to mitigate this?

Accepted Answer

Implement input validation and output filtering with safety attributes. Input validation and output filtering with safety attributes directly address security vulnerabilities by sanitizing user inputs and filtering model outputs for harmful content. The Vertex AI PaLM API provides safety attribute scores (e.g., toxicity, harassment) that allow developers to programmatically block or flag responses that exceed defined thresholds, reducing the risk of generating insecure code snippets.

Answer

Disable safety filters to allow more output

Answer

Increase the max output tokens

Answer

Set safety settings to block all categories

Question 12

Which Google AI model was the first to demonstrate that transformers could be pre-trained bidirectionally on a large corpus, leading to major improvements in language understanding?

Accepted Answer

BERT. BERT (Bidirectional Encoder Representations from Transformers) was the first model to demonstrate that transformers could be pre-trained bidirectionally on a large corpus (BooksCorpus and English Wikipedia). By using a masked language model (MLM) objective, BERT conditions on both left and right context simultaneously, unlike previous unidirectional models, leading to significant improvements on 11 NLP benchmarks at its release.

Answer

GPT-3

Answer

AlphaGo

Answer

Transformer (the paper)

Question 13

A company is considering whether to use Vertex AI's Generative AI Studio. Which TWO are benefits?

Accepted Answer

It integrates seamlessly with Vertex AI Pipelines for MLOps. Vertex AI Generative AI Studio is designed to work natively with Vertex AI Pipelines, enabling users to incorporate generative models into end-to-end MLOps workflows for automation, monitoring, and retraining. This integration allows seamless orchestration of prompt tuning, model evaluation, and deployment within the same managed environment, reducing operational overhead.

Answer

It is always cheaper than using third-party APIs

Answer

It generates outputs that are always more accurate than custom models

Answer

It requires no coding or machine learning expertise to use

Question 14

A software company wants to provide users with a clear understanding of when and why their AI system may produce incorrect answers. Which tool from the Responsible AI toolkit should they use to communicate model limitations?

Accepted Answer

Model Cards. Model Cards are designed to communicate model performance, intended use, and limitations to stakeholders in a standardized format.

Answer

People + AI Guidebook

Answer

PAIR Explorables

Answer

Datasheets for Datasets

Question 15

An enterprise is deploying a customer-facing chatbot using a foundation model on Vertex AI. They need to ensure the model does not produce toxic outputs. Which combination of settings and features should they implement?

Accepted Answer

Configure safety filters and safety settings in the model deployment to block harmful categories. Safety filters and safety settings in Vertex AI model deployment are the direct mechanism to block harmful categories of output at inference time. These settings allow administrators to define thresholds for categories like toxicity, harassment, and hate speech, ensuring the model refuses to generate prohibited content without requiring retraining or post-hoc monitoring.

Answer

Use Reinforcement Learning from Human Feedback (RLHF) during fine-tuning

Answer

Enable Vertex AI Model Monitoring and set up alerts for toxic outputs

Answer

Reduce the temperature to 0.0 and increase top-k to 50