Generative AI Leader Generative AI Concepts and Technologies Practice Test 5 — 15 Questions

Question 1

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 2

Which Google Cloud service is specifically designed for generating code from natural language descriptions?

Accepted Answer

Codey. Codey is the correct answer because it is Google Cloud's foundational code generation model, specifically designed to accept natural language prompts and produce source code in various programming languages. It powers features like code completion, code generation, and chat-based code assistance within Vertex AI, directly addressing the requirement of generating code from natural language descriptions.

Answer

Imagen

Answer

Chirp

Answer

Gemini

Question 3

A researcher wants to adapt a large language model for a specialized medical terminology domain without retraining the entire model. Which fine-tuning method is MOST parameter-efficient?

Accepted Answer

Adapter-based fine-tuning using LoRA. LoRA (Low-Rank Adaptation) is the most parameter-efficient fine-tuning method because it injects trainable low-rank matrices into the transformer layers, updating only a tiny fraction (often <1%) of the model's parameters while keeping the original weights frozen. This allows the model to adapt to specialized medical terminology without the memory and compute cost of full fine-tuning, making it ideal for domain adaptation with limited resources.

Answer

In-context learning with 50 examples

Answer

RLHF (Reinforcement Learning from Human Feedback)

Answer

Full supervised fine-tuning of all model weights

Question 4

An enterprise deploys a generative AI chatbot that must comply with GDPR right to deletion. Users can request deletion of their personal data. The chatbot uses a RAG pipeline with a vector database. What is the MOST effective way to handle deletion requests?

Accepted Answer

Delete the user's documents from the vector index and original storage, then rebuild the index. GDPR requires that personal data be erased when requested. In a RAG system, the source documents must be deleted from the vector store and the original storage. Fine-tuning or model training is not the right approach.

Answer

Update the user's records in the vector index with anonymized placeholders

Answer

Add a filter to the chat application to block the user's name from appearing in responses

Answer

Retrain the LLM from scratch without the user's data

Question 5

A startup wants to generate realistic product videos from text descriptions for social media ads. Which Google Cloud service should they use?

Accepted Answer

Veo. Veo is Google Cloud's advanced video generation model that can create high-quality, realistic videos from text prompts, making it the ideal choice for generating product videos for social media ads. Unlike other services, Veo is specifically designed for video synthesis, offering capabilities like style control and cinematic effects directly from text descriptions.

Answer

Imagen

Answer

Gemini Pro Vision

Answer

Codey

Question 6

What is the key advantage of using vector search for retrieval in a RAG system compared to keyword search?

Accepted Answer

Vector search can find conceptually similar documents even without exact keyword matches. Vector search in a RAG system encodes documents and queries into dense vector embeddings using a foundation model, then retrieves documents based on semantic similarity in the embedding space. This allows it to find conceptually related documents even when they share no exact keywords with the query, overcoming the lexical gap that limits keyword search.

Answer

Vector search eliminates the need for a foundation model

Answer

Vector search is faster than keyword search

Answer

Vector search requires no preprocessing of documents

Question 7

A team is building a multilingual customer support chatbot using Gemini. They notice that for low-resource languages, the model frequently produces grammatically incorrect responses. Which strategy would MOST effectively improve quality for these languages without sacrificing latency?

Accepted Answer

Use few-shot prompting with high-quality examples in the target language. Few-shot prompting with high-quality examples in the target language directly conditions the model to produce grammatically correct responses for low-resource languages without adding inference latency. This approach leverages Gemini's in-context learning capability, allowing it to adapt to the target language's grammar patterns without requiring additional model training or translation round-trips.

Answer

Implement a separate translation step: translate user input to English, generate response in English, then translate back

Answer

Fine-tune the model on a large parallel corpus for each low-resource language

Answer

Increase temperature to 1.5 to encourage more diverse grammar

Question 8

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 9

A company uses a Gemini 1.5 Pro model with a 1 million token context window. They want to process a large 500-page PDF for Q&A. What is the MAIN advantage of using the long context window over a RAG approach?

Accepted Answer

Simpler architecture with no need for chunking or retrieval systems. Option A is correct because the primary advantage of using a 1 million token context window is architectural simplicity. By ingesting the entire 500-page PDF as a single prompt, the company eliminates the need for document chunking, embedding generation, and a retrieval system (RAG). This reduces system complexity, maintenance overhead, and potential failure points, as the model can directly attend to all content in one pass.

Answer

Lower cost per API call

Answer

Faster inference speed

Answer

Higher accuracy for all queries

Question 10

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, more efficient model variant designed for lower latency and higher throughput compared to Gemini Pro. By switching to Flash, the engineer reduces the computational overhead per request, directly decreasing response time for the chatbot.

Answer

Enable streaming mode

Answer

Fine-tune the model on the specific task

Answer

Increase the temperature to 1.0

Question 11

A financial institution wants to deploy a generative AI system for automated report generation. They require that the model does NOT expose sensitive information from its training data and that outputs are factually accurate. Which THREE techniques should they combine?

Accepted Answer

Use Retrieval-Augmented Generation (RAG) to ground outputs in verified documents. RAG grounds outputs in verified sources, RLHF can reduce hallucination and harmful outputs, and differential privacy during training prevents memorization of sensitive data.

Answer

Use a larger model with more parameters

Answer

Increase the temperature to 2.0 for creativity

Question 12

A team is deciding between using fine-tuning and in-context learning for a document classification task. They have 500 labeled examples and need low latency. Which TWO statements are accurate?

Accepted Answer

Fine-tuning generally improves accuracy more than in-context learning when sufficient labeled data is available. Option D is correct because fine-tuning updates the model's weights on a labeled dataset, which generally leads to higher accuracy than in-context learning when sufficient labeled data (like 500 examples) is available. In-context learning relies on the model's pre-existing knowledge and a few examples in the prompt, which often yields lower accuracy for complex classification tasks.

Answer

In-context learning always has lower latency than fine-tuning

Answer

In-context learning can only be used with models that have a context window smaller than 1000 tokens

Answer

Fine-tuning eliminates the need for a validation dataset

Question 13

A company wants to build a customer service chatbot that answers questions about their internal policy documents. The documents are updated monthly, and the team cannot afford to retrain a model each time. Which approach is MOST appropriate?

Accepted Answer

Use Retrieval-Augmented Generation (RAG) with the policy documents indexed in a vector store. RAG (Retrieval-Augmented Generation) allows the LLM to retrieve relevant document sections at inference time, so knowledge stays current without retraining. The other options either require expensive retraining for each update or lack document grounding.

Answer

Train a custom model from scratch on the policy documents each month

Answer

Fine-tune a base LLM on the policy documents monthly

Answer

Use a larger foundation model with a longer context window and paste all documents into each prompt

Question 14

What is the primary purpose of the temperature parameter when generating text with an LLM?

Accepted Answer

It controls the randomness or creativity of the output. Temperature controls the randomness of token selection. Higher temperature increases creativity, lower temperature makes outputs more deterministic.

Answer

It sets the maximum number of tokens in the response

Answer

It determines the number of most likely tokens considered at each step

Answer

It sets the cumulative probability threshold for token selection

Question 15

A data scientist is using Vertex AI to fine-tune a Gemini model for a specialized legal document summarization task. They have a small set of labeled examples (200 pairs). Which fine-tuning method is MOST cost-effective and likely to perform well?

Accepted Answer

Adapter-based fine-tuning (e.g., LoRA). Adapter-based fine-tuning (like LoRA) updates only a small fraction of parameters, making it efficient with small datasets and low cost, while still adapting the model to the task.

Answer

Full fine-tuning of all model parameters

Answer

Training a small custom model from scratch

Answer

Prompt engineering with few-shot examples only