Generative AI Leader Practice Test 47 — 15 Questions

Question 1

A financial institution deploys a chatbot using Gemini Pro in Vertex AI. Compliance requires logging all user inputs and model outputs for audit. Which approach meets this requirement?

Accepted Answer

Enable Vertex AI Endpoint request-response logging. Vertex AI Endpoint request-response logging captures both the user's input prompt and the model's generated output, which is precisely what compliance auditing requires. This feature logs the exact payloads sent to and received from the deployed model, ensuring a complete audit trail without additional configuration.

Answer

Capture logs via Cloud Monitoring

Answer

Use Cloud Logging sink with a filter for Vertex AI requests

Answer

Enable Vertex AI Model Registry logging

Question 2

Refer to the exhibit. A data scientist runs the gcloud command and sees the model listed. However, when they try to deploy the model to an endpoint, they get an error: 'Model is not deployable'. What is the most likely reason?

Accepted Answer

The model was imported from a custom container but without a serving specification or artifact.. A model imported from a custom container must include a serving specification (e.g., a `predict` route) and an artifact (e.g., a saved model file) to be deployable. Without these, Vertex AI cannot determine how to serve predictions, resulting in the 'Model is not deployable' error. The `gcloud` command listing the model only confirms its registration, not its readiness for deployment.

Answer

The model is still in training and not yet ready.

Answer

The model does not have the correct IAM permissions assigned to the deployment service account.

Answer

The region for the endpoint is different from the model's region.

Question 3

A company wants to use generative AI to summarize customer support tickets. Which Google Cloud tool is best suited for this task?

Accepted Answer

Vertex AI Text Generation (Gemini). Vertex AI Text Generation (Gemini) is the correct choice because it is a generative AI service specifically designed for natural language understanding and generation tasks, such as summarizing customer support tickets. Gemini models can process long-form text and produce concise, coherent summaries by leveraging transformer-based architectures fine-tuned for instruction-following and text completion. This makes it ideal for extracting key information from support conversations and generating actionable summaries.

Answer

Dialogflow CX

Answer

Document AI

Answer

AutoML Tables

Question 4

A startup is building a customer service chatbot that generates responses in real-time. They want the model to have up-to-date information on the latest product catalog but cannot afford frequent fine-tuning. Which technique should they use to inject current data into the model without retraining?

Accepted Answer

Use retrieval-augmented generation (RAG) to fetch relevant documents from a vector database at inference time.. Retrieval-Augmented Generation (RAG) is the correct technique because it allows the chatbot to fetch the most current product catalog entries from an external vector database at inference time, without requiring any model retraining. This keeps responses grounded in up-to-date information while avoiding the cost and latency of frequent fine-tuning.

Answer

Rely on the model's zero-shot capabilities to infer product details.

Answer

Craft detailed system prompts that include the entire product catalog in the prompt.

Answer

Fine-tune the base model weekly on the latest product catalog.

Question 5

Refer to the exhibit. A developer sees this error when trying to call a Vertex AI endpoint for online prediction. What permission does the requesting identity need to be granted?

Accepted Answer

aiplatform.endpoints.predict. The error occurs when calling a Vertex AI endpoint for online prediction, which requires the `aiplatform.endpoints.predict` permission. This permission is specifically scoped to the endpoint resource, allowing the identity to send prediction requests to a deployed model endpoint. The correct IAM role binding must include this permission for the requesting identity to successfully invoke the endpoint.

Answer

aiplatform.prediction.predict

Answer

aiplatform.endpoints.use

Answer

aiplatform.models.predict

Question 6

A gen AI application produces hallucinations (factually incorrect outputs). Which mitigation strategy is LEAST effective?

Accepted Answer

Increasing model temperature. Increasing model temperature makes the model more random and creative, which directly increases the likelihood of hallucinations. It does not constrain or ground the output in factual data, making it the least effective mitigation strategy among the options.

Answer

Using prompt templates with constraints

Answer

Using grounding with a knowledge base

Answer

Implementing retrieval-augmented generation

Question 7

A data scientist is using Vertex AI to build a question-answering system. They want to use the Responsible AI toolkit to document and communicate the model's characteristics. Which TWO tools from the toolkit are MOST relevant? (Select two.)

Accepted Answer

Model Cards. Model Cards (C) are the Responsible AI tool designed to document a model's characteristics—its intended use, performance metrics, limitations, and ethical considerations—which directly supports the goal of documenting and communicating model characteristics. Datasheets for Datasets (D) complement this by documenting the dataset's provenance, composition, and collection process, providing the data-side documentation needed for responsible communication of the system. Cloud DLP (A) is a data loss prevention and de-identification service, not a documentation tool. TensorBoard (B) is a visualization tool for training metrics, and Vertex AI Workbench (E) is a managed notebook environment for development—neither is part of the Responsible AI documentation toolkit.

Answer

Cloud DLP

Answer

TensorBoard

Answer

Vertex AI Workbench

Question 8

A healthcare company is deploying a GenAI-powered report generation system that processes patient summaries. They need the output to be structured JSON for downstream ingestion. The team is using Vertex AI Studio prompt design. Which approach best ensures consistent structured output?

Accepted Answer

Define the JSON schema in the system instruction and provide one few-shot example of the expected JSON. Defining the JSON schema in the system instruction and providing a few-shot example directly constrains the model's output format at inference time, leveraging Vertex AI's instruction-following capabilities. This approach ensures the model generates valid JSON consistently without post-processing, as the schema acts as a structural template that the model learns to replicate.

Answer

Ask the model to output plain text and parse it with a regex after generation

Answer

Use a lower temperature (0.0) and hope the model outputs valid JSON

Answer

Generate output in Markdown and convert to JSON using a secondary script

Question 9

A company wants to use AI to make hiring decisions. They are concerned about bias against certain demographic groups. According to Google's AI Principles, which approach is MOST aligned?

Accepted Answer

Evaluate the model using diverse test sets and adjust if bias is found. The principle 'avoid creating or reinforcing unfair bias' requires proactive identification and mitigation. Evaluating the model on diverse test sets is a standard way to detect and address bias before deployment.

Answer

Pre-train the model on a dataset that is balanced across all demographics

Answer

Blind the model to demographic features to ensure fairness

Answer

Only use the model for initial resume screening, with final decisions by humans

Question 10

A company plans to commercially use images generated by a text-to-image model. What should they check to avoid copyright issues?

Accepted Answer

The training data provenance and the model's license terms. Copyright issues arise from the training data and the model's license. The training data provenance determines if the model was trained on copyrighted works without permission, and the license terms specify whether commercial use is allowed. Without checking both, the company risks infringing on original creators' rights.

Answer

The model's accuracy on standard benchmarks

Answer

The model's latency and throughput

Answer

The model's bias evaluation results

Question 11

A machine learning engineer is evaluating a generative AI model for bias. They have a diverse test set covering gender, race, and age groups. Which metric would best indicate if the model's performance is systematically worse for certain demographic groups?

Accepted Answer

Equalized odds across demographic groups. Equalized odds measures whether a model's predictions have equal false positive/negative rates across groups. The other options either measure different aspects or are not specific to fairness.

Answer

Model perplexity on held-out data

Answer

Overall accuracy on the test set

Answer

Area under the ROC curve (AUC)

Question 12

A global company deploys a generative AI chatbot in the European Union. They must ensure compliance with GDPR regarding user data used for fine-tuning the model. What is the MOST important requirement they must fulfill?

Accepted Answer

Obtain explicit consent from data subjects for using their data in fine-tuning. Under GDPR, using personal data for fine-tuning a generative AI model constitutes a new processing purpose that requires a lawful basis. Explicit consent (Article 7 and Article 9) is the most robust basis when relying on consent, as it must be freely given, specific, informed, and unambiguous. Without explicit consent, the company risks violating data minimization and purpose limitation principles, even if other anonymization or storage measures are applied.

Answer

Use only synthetic data for fine-tuning to avoid GDPR issues

Answer

Store all fine-tuned models only on US-based servers

Answer

Anonymize all data before fine-tuning, regardless of consent

Question 13

A company wants to use a generative AI model to create marketing content that will be used commercially. They are concerned about copyright infringement from the model's training data. Which step should they take to mitigate this risk?

Accepted Answer

Review the model's training data provenance and ensure the model is trained on properly licensed or public domain data. Reviewing the training data provenance and using models with appropriate licenses is critical. The other options either do not address the root cause or are insufficient.

Answer

Use SynthID to watermark the content

Answer

Add a disclaimer that content is AI-generated

Answer

Use only open-source models

Question 14

What is the primary purpose of Google's Content Safety filters in Vertex AI?

Accepted Answer

To block generated content that contains hate speech, violence, or sexually explicit material. Google's Content Safety filters in Vertex AI are designed to block generated content that violates safety policies, specifically targeting hate speech, violence, and sexually explicit material. This is a core component of responsible AI deployment, ensuring that model outputs adhere to ethical guidelines and legal requirements. The filters operate by analyzing the generated text or images against predefined safety categories, not by assessing data quality or source curation.

Answer

To filter out low-quality training data

Answer

To ensure the model only generates content from a curated set of sources

Answer

To improve the model's accuracy on safe content

Question 15

A startup is building a generative AI legal document assistant for small law firms. They want to ensure that the model's outputs are accurate and can be traced back to specific legal statutes. Which approach best supports this requirement?

Accepted Answer

Use a RAG architecture that retrieves relevant statutes and includes them as citations in the model's response. Retrieval-Augmented Generation (RAG) architecture retrieves specific legal statutes from a trusted external knowledge base and includes them as citations in the model's response. This ensures both accuracy (by grounding outputs in verifiable sources) and traceability (by providing direct references to the statutes used). Fine-tuning alone cannot guarantee that the model will cite specific statutes correctly, as it may hallucinate or misremember legal references.

Answer

Fine-tune the model on a large corpus of legal documents

Answer

Apply a high temperature setting to encourage diverse outputs

Answer

Use a model larger than 70B parameters