Microsoft - AI-901: Microsoft Azure AI Fundamentals (beta)
Sample Questions
Question: 60
Measured Skill: Identify AI concepts and responsibilities (40–45%)
Select the answer that correctly completes the sentence.
| A | Evaluating model outcomes across demographic groups to reduce bias is an example of the Microsoft responsible AI principle of accountability. |
| B | Evaluating model outcomes across demographic groups to reduce bias is an example of the Microsoft responsible AI principle of fairness. |
| C | Evaluating model outcomes across demographic groups to reduce bias is an example of the Microsoft responsible AI principle of privacy and security. |
| D | Evaluating model outcomes across demographic groups to reduce bias is an example of the Microsoft responsible AI principle of transparency. |
Correct answer: BExplanation:
Microsoft developed a Responsible AI Standard. It's a framework for building AI systems according to six principles: fairness, reliability and safety, privacy and security, inclusiveness, transparency, and accountability. For Microsoft, these principles are the cornerstone of a responsible and trustworthy approach to AI, especially as intelligent technology becomes more prevalent in products and services that people use every day.
Fairness and inclusiveness
AI systems should treat everyone fairly and avoid affecting similarly situated groups of people in different ways. For example, when AI systems provide guidance on medical treatment, loan applications, or employment, they should make the same recommendations to everyone who has similar symptoms, financial circumstances, or professional qualifications.
Reliability and safety
To build trust, it's critical that AI systems operate reliably, safely, and consistently. These systems should be able to operate as they were originally designed, respond safely to unanticipated conditions, and resist harmful manipulation. How they behave and the variety of conditions they can handle reflect the range of situations and circumstances that developers anticipated during design and testing.
Transparency
When AI systems help inform decisions that have tremendous impacts on people's lives, it's critical that people understand how those decisions were made. For example, a bank might use an AI system to decide whether a person is creditworthy. A company might use an AI system to determine the most qualified candidates to hire.
A crucial part of transparency is interpretability: the useful explanation of the behavior of AI systems and their components. Improving interpretability requires stakeholders to comprehend how and why AI systems function the way they do. The stakeholders can then identify potential performance issues, fairness issues, exclusionary practices, or unintended outcomes.
Privacy and security
As AI becomes more prevalent, protecting privacy and securing personal and business information are becoming more important and complex. With AI, privacy and data security require close attention because access to data is essential for AI systems to make accurate and informed predictions and decisions about people. AI systems must comply with privacy laws that:
- Require transparency about the collection, use, and storage of data.
- Mandate that consumers have appropriate controls to choose how their data is used.
Accountability
The people who design and deploy AI systems must be accountable for how their systems operate. Organizations should draw upon industry standards to develop accountability norms. These norms can ensure that AI systems aren't the final authority on any decision that affects people's lives. They can also ensure that humans maintain meaningful control over otherwise highly autonomous AI systems.
Reference: What is Responsible AI?
Question: 61
Measured Skill: Identify AI concepts and responsibilities (40–45%)
For each of the following statements, select Yes if the statement is true. Otherwise, select No.
(NOTE: Each correct selection is worth one point.)
| A | Evaluators in Microsoft Foundry replace the need for configuring token limits: Yes
Evaluators in Microsoft Foundry can assess the quality and safety of responses generated by a generative AI model: Yes
Evaluators in Microsoft Foundry can retrain a deployed generative AI model automatically when quality issues are detected: Yes
|
| B | Evaluators in Microsoft Foundry replace the need for configuring token limits: Yes
Evaluators in Microsoft Foundry can assess the quality and safety of responses generated by a generative AI model: Yes
Evaluators in Microsoft Foundry can retrain a deployed generative AI model automatically when quality issues are detected: No |
| C | Evaluators in Microsoft Foundry replace the need for configuring token limits: No
Evaluators in Microsoft Foundry can assess the quality and safety of responses generated by a generative AI model: Yes
Evaluators in Microsoft Foundry can retrain a deployed generative AI model automatically when quality issues are detected: No |
| D | Evaluators in Microsoft Foundry replace the need for configuring token limits: No
Evaluators in Microsoft Foundry can assess the quality and safety of responses generated by a generative AI model: Yes
Evaluators in Microsoft Foundry can retrain a deployed generative AI model automatically when quality issues are detected: Yes |
| E | Evaluators in Microsoft Foundry replace the need for configuring token limits: No
Evaluators in Microsoft Foundry can assess the quality and safety of responses generated by a generative AI model: No
Evaluators in Microsoft Foundry can retrain a deployed generative AI model automatically when quality issues are detected: Yes |
| F | Evaluators in Microsoft Foundry replace the need for configuring token limits: No
Evaluators in Microsoft Foundry can assess the quality and safety of responses generated by a generative AI model: No
Evaluators in Microsoft Foundry can retrain a deployed generative AI model automatically when quality issues are detected: No |
Correct answer: CExplanation:
In Microsoft Foundry (Azure AI Foundry), evaluators are used to automatically measure and improve the quality of AI model outputs. They are part of the model evaluation workflow rather than something that directly controls model behavior like token limits or training.
Evaluators take agent messages as input and output binary Pass/Fail scores (or scaled scores converted to binary scores based on thresholds). These evaluators support two best practices for agent evaluation:
- System evaluation - to examine the end-to-end outcomes of the agentic system.
- Process evaluation - to verify the step-by-step execution to achieve the outcomes.
System evaluation
System evaluation examines the quality of the final outcome of your agentic workflow. These evaluators are applicable to single agents and, in multi-agent systems, to the main orchestrator or the final agent responsible for task completion:
- Task Completion - Did the agent fully complete the requested task?
- Task Adherence - Did the agent follow the rules and constraints in its instructions?
- Task Navigation Efficiency - Did the agent perform the expected steps efficiently?
- Intent Resolution - Did the agent correctly identify and address user intentions?
Specifically, for textual outputs from agents, you can also apply RAG quality evaluators such as Relevance and Groundedness that take agentic inputs to assess the final response quality.
Process evaluation
Process evaluation examines the quality and efficiency of each step in your agentic workflow. These evaluators focus on the tool calls executed in a system to complete tasks:
- Tool Call Accuracy - Did the agent make the right tool calls with correct parameters without redundancy?
- Tool Selection - Did the agent select the correct and necessary tools?
- Tool Input Accuracy - Did the agent provide correct parameters for tool calls?
- Tool Output Utilization - Did the agent correctly use tool call results in its reasoning and final response?
- Tool Call Success - Did the tool calls succeed without technical errors?
References:
Agent evaluators
Built-in evaluators reference
Question: 62
Measured Skill: Identify AI concepts and responsibilities (40–45%)
For each of the following statements, select Yes if the statement is true. Otherwise, select No.
(NOTE: Each correct selection is worth one point.)
| A | Generating a response to a user prompt occurs during the inference stage: Yes
A generative AI model generates responses by copying stored documents directly from the model’s training data: Yes
A generative AI model produces output by predicting the next token based on patterns learned from the model’s training data: Yes |
| B | Generating a response to a user prompt occurs during the inference stage: Yes
A generative AI model generates responses by copying stored documents directly from the model’s training data: Yes
A generative AI model produces output by predicting the next token based on patterns learned from the model’s training data: No |
| C | Generating a response to a user prompt occurs during the inference stage: Yes
A generative AI model generates responses by copying stored documents directly from the model’s training data: No
A generative AI model produces output by predicting the next token based on patterns learned from the model’s training data: Yes |
| D | Generating a response to a user prompt occurs during the inference stage: No
A generative AI model generates responses by copying stored documents directly from the model’s training data: Yes
A generative AI model produces output by predicting the next token based on patterns learned from the model’s training data: No |
| E | Generating a response to a user prompt occurs during the inference stage: No
A generative AI model generates responses by copying stored documents directly from the model’s training data: No
A generative AI model produces output by predicting the next token based on patterns learned from the model’s training data: Yes |
| F | Generating a response to a user prompt occurs during the inference stage: No
A generative AI model generates responses by copying stored documents directly from the model’s training data: No
A generative AI model produces output by predicting the next token based on patterns learned from the model’s training data: No |
Correct answer: CExplanation:
Generative AI refers to a class of AI models, such as the GPT series or Llama, that analyzes large amounts of data and generates new content, including text, images, and code, that mirrors human expression—redefining our relationship to technology.
The inference stage is the phase where a trained AI model is used to generate responses or predictions based on new input. In the training stage, a model learns patterns from large datasets and adjusts internal weights (parameters). In the inference stage, a model receives a new prompt/input and produces an output/response.
Generative AI models do not copy or retrieve stored documents. They generate responses by learning patterns, not memorizing content.
next-token prediction is the core mechanism of generative AI.
References:
How does generative AI work?
What is generative AI?
Question: 63
Measured Skill: Implement AI solutions by using Microsoft Foundry (55–60%)
For each of the following statements, select Yes if the statement is true. Otherwise, select No.
(NOTE: Each correct selection is worth one point.)
| A | The Temperature parameter can be set before deploying a model: Yes
During inference, the model name is used to route requests to a specific deployment: Yes
After a model is deployed, both code and testing tools can be used to interact with the model: Yes |
| B | The Temperature parameter can be set before deploying a model: Yes
During inference, the model name is used to route requests to a specific deployment: Yes
After a model is deployed, both code and testing tools can be used to interact with the model: No |
| C | The Temperature parameter can be set before deploying a model: Yes
During inference, the model name is used to route requests to a specific deployment: No
After a model is deployed, both code and testing tools can be used to interact with the model: No |
| D | The Temperature parameter can be set before deploying a model: Yes
During inference, the model name is used to route requests to a specific deployment: No
After a model is deployed, both code and testing tools can be used to interact with the model: Yes |
| E | The Temperature parameter can be set before deploying a model: No
During inference, the model name is used to route requests to a specific deployment: No
After a model is deployed, both code and testing tools can be used to interact with the model: Yes |
| F | The Temperature parameter can be set before deploying a model: No
During inference, the model name is used to route requests to a specific deployment: No
After a model is deployed, both code and testing tools can be used to interact with the model: No |
Correct answer: DExplanation:
In playgrounds and tools, you can configure parameters like temperature when working with models. The playground can be used for testing custom settings configuration before deployment. Temperature is a configurable parameter, before or during inference.
Requests are routed using the deployment name, not the raw model name. You pass your deployment name to the model parameter.
After you deploy a model you can interact with it using both UI tools such as the playground and code, such as SDKs and the response API. Playgrounds enable testing, experimentation, and API exploration.
References:
Microsoft Foundry Playgrounds
Auto and direct model routing with the Responses API
Question: 64
Measured Skill: Implement AI solutions by using Microsoft Foundry (55–60%)
You are developing an application that extracts structured information from different types of content by using Azure Content Understanding in Foundry Tools.
You need to extract scanned invoices in the PDF format and voicemail recordings in the WAV format.
Which type of analyzer should you use for each content type?
(To answer, drag the appropriate analyzer types to the correct content types. Each analyzer type may be used once, more than once, or not at all. NOTE: Each correct selection is worth one point.)
| A | Scanned invoices: Video analyzer
Voicemail recordings: Image analyzer |
| B | Scanned invoices: Document analyzer
Voicemail recordings: Image analyzer |
| C | Scanned invoices: Audio analyzer
Voicemail recordings: Document analyzer |
| D | Scanned invoices: Image analyzer
Voicemail recordings: Audio analyzer |
| E | Scanned invoices: Image analyzer
Voicemail recordings: Video analyzer |
| F | Scanned invoices: Document analyzer
Voicemail recordings: Audio analyzer |
Correct answer: FExplanation:
An analyzer in Azure Content Understanding in Foundry Tools is a configurable processing unit that defines how your content is analyzed and what information is extracted.
An analyzer defines:
- What type of content to process (documents, images, audio, or video)
- What elements to extract (text, layout, tables, fields, transcripts)
- How to structure the output (markdown, JSON fields, segments)
- Which AI models to use for processing
Analyzers are the core building blocks of Content Understanding. They combine content extraction, AI-powered analysis, and structured data output into a single, reusable configuration. You can use prebuilt analyzers for common scenarios or create custom analyzers tailored to your specific needs.
Base analyzers
Base analyzers provide fundamental content processing capabilities specific to a content type. Use them primarily as a parent to inherit from when creating custom analyzers. When you create a custom analyzer, include one of these base analyzers by using the baseAnalyzerId property.
prebuilt-audio - Base audio processingprebuilt-document - Base document processingprebuilt-image - Base image processingprebuilt-video - Base video processing
References:
What is a Content Understanding analyzer?
Prebuilt analyzers in Azure Content Understanding in Foundry Tools