Online Access Free AB-731 Exam Questions

You have a large language model (LLM) that was pretrained by using public data.
You want the LLM to generate responses that incorporate frequently updated proprietary content, such as internal documents and knowledge bases.
You need to recommend a solution to ensure that the LLM uses the most current information without retraining the model.
What should you include in the recommendation?

• A.prompt engineering
• B.content filtering
• C.Retrieval Augmented Generation (RAG)
• D.model fine-tuning

Your company sells hiking and camping gear online.
You need a generative AI solution that can interact with customers and ask questions about their needs.
What should you include in the solution?

• A.a chatbot
• B.computer vision
• C.a recommendation engine
• D.predictive AI

Hotspot Question
Select the answer that correctly completes the sentence.

Correct Answer:

Hotspot Question
For each of the following statements, select Yes if the statement is true. Otherwise, select No.
NOTE: Each correct selection is worth one point.

Correct Answer:

Explanation:
Box 1: Yes
Yes - Content filtering controls can prevent AI-generated responses from exposing confidential and sensitive information.
Content filtering controls are a critical safeguard for preventing AI-generated responses from inadvertently exposing confidential or sensitive information. These controls operate as
"guardrails" by inspecting data at both the input (prompt) and output (completion) stages.
Box 2: Yes
Yes - AI-generated content can unintentionally reveal sensitive information if the generative AI model has access to unsecured data source.
Unsecured data sources significantly increase the risk of AI-generated content exposing sensitive information through several mechanisms. Because generative AI models are "data-hungry" and often lack transparent data boundaries, they can absorb and later reproduce confidential data they were exposed to during training or operation.
Box 3: No
No - To prevent data exposure, only the prompts used by users must be protected by using policies.
While protecting user prompts is a critical first step, it is not sufficient on its own to prevent data exposure. Relying solely on prompt policies leaves significant gaps where sensitive data can still leak through other vectors.
Reference:
https://noma.security/resources/ai-application-security/
https://pacific.ai/managing-privacy-risks-llm-guidance/
https://www.paloaltonetworks.com/blog/network-security/securing-the-future-by-protecting-sensitive-data-in-ai-systems

Hotspot Question
For each of the following statements, select Yes if the statement is true. Otherwise, select No.
NOTE: Each correct selection is worth one point.

Correct Answer:

Explanation:
Box 1: Yes
Yes - Azure Vision in Foundry Tools can extract and analyze key phrases from PDF files.
Azure Document Intelligence (formerly part of Azure AI Services, now integrated into Azure AI Foundry Tools) can extract text, tables, and structures from PDF files.
While Azure Vision specifically handles Optical Character Recognition (OCR) for scanning text in images and documents, the combined capabilities within the Foundry ecosystem-particularly using Document Intelligence-allow for the extraction of structured data and, when combined with Azure Language services, the identification of key phrases and semantic information.
Box 2: No
No - Azure Vision in Foundry Tools can generate images based on natural language descriptions.
Azure Vision in Foundry Tools is designed for analyzing existing visual content rather than generating new images from scratch.
While the "Foundry" platform does offer image generation, it is typically handled by a separate image generation tool (currently in preview) that uses models like DALL-E Box 3: Yes Yes - Azure Document Intelligence in Foundry Tools can be used to automate the processing of invoices and credit notes.
Azure Document Intelligence in Foundry Tools (formerly part of Azure AI Services) is designed to automate the processing of invoices and credit notes, transforming unstructured documents into structured, actionable data within workflows.
It uses machine learning and Optical Character Recognition (OCR) to extract key fields (such as vendor name, invoice date, amounts, and tax information) and line items.
Reference:
https://azure.microsoft.com/en-in/products/ai-foundry/tools/document-intelligence
https://learn.microsoft.com/en-us/azure/ai-services/computer-vision/concept-describing-images
https://azure.microsoft.com/en-in/products/ai-foundry/tools/document-intelligence