wayground logo

Free Printable Worksheets

Font size

S
M
L
XL
Worksheets

Generative AI/Fairness and Ethics

Total questions: 64

Worksheet time: 32mins

Name
Class
Date
1.

Some mitigation techniques you can use at the metaprompt and grounding layer include:

a)

Clarifying user intent and providing context

b)

Ignoring user input

c)

Generating random responses

d)

Avoiding any user feedback

2.

A Large Language Model (LLM) is:

a)

a type of artificial intelligence that processes and generates human-like text

b)

a device used for storing large amounts of data

c)

a programming language for building web applications

d)

a tool for creating digital images

3.

The first step in training a transformer model is:

a)

Tokenizing the input data

b)

Evaluating the model

c)

Deploying the model

d)

Fine-tuning the model

4.

LLMs are based on which type of architecture?

a)

Convolutional Neural Network

b)

Recurrent Neural Network

c)

Transformer

d)

Decision Tree

5.

An encoder block is:

a)

a component that processes input data and extracts features in a neural network

b)

a device that converts analog signals to digital signals

c)

a hardware component used for data storage

d)

a type of memory used in computers

6.

A decoder block is:

a)

a component that converts encoded data back to its original form

b)

a device that encodes information for transmission

c)

a block that stores data temporarily

d)

a circuit that amplifies signals

7.

What best describes Generative AI?

a)

AI that analyzes and summarizes existing data

b)

AI that creates new content such as text, images, or code

c)

AI that only classifies data into categories

d)

AI that detects anomalies in data

8.

Which Azure service provides access to pre-trained generative language and image models?

a)

Azure Machine Learning

b)

Azure AI Studio

c)

Azure Cognitive Search

d)

Azure Data Factory

9.

What is the primary advantage of using foundation models?

a)

They require manual training from scratch for each task

b)

They can be fine-tuned for many tasks with less data

c)

They only work for image generation

d)

They eliminate all ethical risks

10.

What is a prompt in generative AI?

a)

A dataset used to train a model

b)

A command that tells a model what to create or output

c)

A data visualization tool

d)

A type of supervised learning

11.

Which of the following is NOT a common generative AI use case?

a)

Text summarization

b)

Image creation

c)

Predicting customer churn

d)

Code completion

12.

What Azure service lets you customize generative AI models safely without writing code?

a)

Azure AI Content Safety

b)

Azure OpenAI Service in Azure AI Studio

c)

Azure Synapse Analytics

d)

Power BI

13.

What is prompt engineering?

a)

Building neural networks manually

b)

Improving a model’s accuracy by tuning weights

c)

Crafting and refining instructions to guide generative model output

d)

Writing Python scripts to deploy models

14.

Large Language Models (LLMs) are primarily trained to:

a)

Identify objects in images

b)

Predict the next word in a sequence

c)

Detect network intrusions

d)

Cluster unlabeled data

15.

What is an example of a multimodal model?

a)

A model that only processes tabular data

b)

A model that handles both text and images as inputs

c)

A model that runs on multiple GPUs

d)

A model with multiple hidden layers

16.

Which Microsoft tool provides guided access to generative AI capabilities for low-code users?

a)

Azure Logic Apps

b)

Power Virtual Agents with Copilot

c)

Microsoft Defender

d)

SQL Server

17.

A marketing team wants to automatically generate product descriptions and social-media posts from a spreadsheet of items. Which Azure service should they use?

a)

Azure Cognitive Search

b)

Azure OpenAI Service

c)

Azure Machine Learning Designer

d)

Azure DevOps

18.

A teacher wants a chatbot that answers students’ questions using only her school’s website content. Which Azure component ensures the model stays within that data?

a)

Azure Cognitive Search indexing + grounding

b)

Azure Content Safety

c)

Azure Cosmos DB

d)

Azure Container Registry

19.

A developer wants to moderate user-generated content for harmful or unsafe language before displaying it. Which service should they add?

a)

Azure AI Content Safety

b)

Azure Monitor

c)

Azure Sentinel

d)

Azure Speech Services

20.

When using Azure OpenAI, which responsibility belongs to the user (not Microsoft)?

a)

Building data centers

b)

Prompt design and responsible deployment

c)

Model pre-training

d)

GPU optimization

21.

Which feature helps prevent data leakage when using enterprise data with generative AI?

a)

Role-based access control and data grounding

b)

Public API endpoints

c)

Anonymous authentication

d)

Hard-coding credentials

22.

An HR department uses generative AI to summarize resumes. Which risk must they consider most?

a)

Overfitting

b)

Data bias and fairness

c)

Under-sampling

d)

Low latency

23.

An e-commerce site uses Azure AI Studio to create product-image variations. Which type of model are they using?

a)

Large Language Model (LLM)

b)

Diffusion or image-generation model

c)

Clustering model

d)

Regression model

24.

Which Azure AI feature supports monitoring for toxicity and bias in generated outputs?

a)

Azure Advisor

b)

Responsible AI dashboard

c)

Azure Resource Monitor

d)

Azure Databricks

25.

Your company wants to integrate generative AI features into a Power App. Which connector or capability allows that?

a)

Power Automate flows

b)

AI Builder with Azure OpenAI integration

c)

Microsoft Fabric

d)

Azure Kubernetes Service

26.

Which principle of Responsible AI relates to giving users clear explanations of how AI decisions are made?

a)

Reliability and safety

b)

Inclusiveness

c)

Transparency

d)

Accountability

27.

Which of the following are examples of generative AI models?

a)

GPT-4

b)

DALL-E 3

c)

ResNet-50

d)

K-Means

28.

Which Azure services can be used to build or customize generative AI solutions?

a)

Azure OpenAI Service

b)

Azure AI Studio

c)

Azure Key Vault

d)

Azure Monitor

29.

Which practices support responsible use of generative AI?

a)

Providing user disclaimers about AI-generated content

b)

Training models only on private data

c)

Evaluating outputs for bias and harm

d)

Allowing unrestricted public prompts

30.

What are common input types for generative AI systems?

a)

Text prompts

b)

Numerical regressions

c)

Images

d)

Hardware addresses

31.

Which statements about Azure OpenAI Service are true?

a)

It provides access to models like GPT-4 and DALL-E through an API

b)

It requires users to build models from scratch

c)

It integrates with Azure AI Content Safety

d)

It can only run on-premises hardware

32.

What is the main goal of responsible AI?

a)

To make AI models run faster

b)

To ensure AI systems are built and used in ways that are ethical, fair, and safe

c)

To increase company profits through automation

d)

To replace human decision-making entirely

33.

Which of the following best describes fairness in AI systems?

a)

AI gives everyone identical outcomes

b)

AI provides equal and appropriate treatment across groups without bias

c)

AI models perform faster for all users

d)

AI avoids human involvement

34.

What is an example of bias in an AI system?

a)

A model predicting the same result every time

b)

A model performing worse for certain demographic groups

c)

A model improving with more data

d)

A model learning to predict new patterns

35.

Which of the following is an ethical risk of AI?

a)

Overfitting a model

b)

Collecting personal data without consent

c)

Optimizing hyperparameters

d)

Using cloud compute resources

36.

Which Microsoft Responsible AI principle focuses on giving users understandable explanations of how AI works?

a)

Reliability and Safety

b)

Transparency

c)

Privacy and Security

d)

Fairness

37.

What does accountability mean in responsible AI?

a)

Only the AI is responsible for its own outcomes

b)

People and organizations remain responsible for AI decisions and impacts

c)

AI systems self-report errors

d)

Accountability is optional in AI

38.

When training a model, which action helps reduce bias?

a)

Collecting diverse and representative datasets

b)

Using fewer training samples

c)

Ignoring outliers

d)

Using a more complex algorithm

39.

What is inclusive design in AI?

a)

Designing AI that only works for specific users

b)

Designing AI systems that consider a variety of abilities, languages, and backgrounds

c)

Removing accessibility features for simplicity

d)

Focusing only on technical accuracy

40.

Which Microsoft service helps identify and mitigate unfairness in models?

a)

Azure AI Content Safety

b)

Responsible AI Dashboard in Azure Machine Learning

c)

Azure Cognitive Search

d)

Azure Synapse Analytics

41.

Why is reliability and safety important in AI?

a)

It reduces server costs

b)

It ensures models behave as intended and handle unexpected inputs safely

c)

It improves the visual design of an app

d)

It helps AI generate more creative outputs

42.

A bank uses an AI model for loan approval. The model denies loans to women more often than men, even with the same income and credit score. Which principle is being violated?

a)

Fairness

b)

Transparency

c)

Inclusiveness

d)

Reliability

43.

A medical AI tool misclassifies darker-skinned patients due to underrepresentation in the dataset. What is the main cause?

a)

Biased training data

b)

Too much model regularization

c)

Model drift

d)

Hyperparameter tuning

44.

A school uses an AI system to predict which students might need academic support. What step promotes ethical use of this AI?

a)

Making predictions public

b)

Explaining the AI’s reasoning to parents and teachers

c)

Using AI without human review

d)

Collecting sensitive data without consent

45.

A company deploys a chatbot that generates offensive responses to users. Which Azure service can help prevent this issue?

a)

Azure AI Content Safety

b)

Azure Kubernetes Service

c)

Azure Monitor

d)

Power Automate

46.

An organization wants to ensure its AI products follow Microsoft’s Responsible AI standards. Which tool provides guided assessments and best practices?

a)

Responsible AI Standard

b)

Azure Key Vault

c)

AI Builder

d)

Azure Sentinel

47.

What is a best practice for ensuring transparency in an AI project?

a)

Documenting data sources and model design choices

b)

Keeping datasets secret

48.

An AI team discovers bias in model predictions. What should they do first?

a)

Delete all data

b)

Identify and analyze the source of the bias in data or labeling

c)

Reduce the model’s accuracy threshold

d)

Ignore it if accuracy is high

49.

When might privacy and security concerns arise in AI systems?

a)

When sensitive personal data is used or shared without proper protection

b)

When using open-source libraries

c)

When models are optimized

d)

When using synthetic data

50.

A company uses an AI hiring tool. To maintain accountability, what should they do?

a)

Allow the AI to make final hiring decisions

b)

Keep humans in the loop to review and approve decisions

c)

Hide AI use from applicants

d)

Automate all HR processes

51.

Which of these is a key feature of Microsoft’s Responsible AI approach?

a)

Encouraging experimentation without limits

b)

Publishing tools and frameworks like the Responsible AI dashboard

c)

Allowing models to evolve independently

d)

Avoiding human oversight

52.

Which are Microsoft’s six principles of Responsible AI?

a)

Fairness

b)

Reliability and Safety

c)

Scalability

d)

Accountability

e)

Efficiency

53.

Which actions improve fairness in an AI solution?

a)

Auditing models for demographic performance differences

b)

Ignoring sensitive attributes

c)

Using diverse data samples

d)

Reducing training data to simplify the model

54.

Which practices promote transparency?

a)

Providing clear explanations of model predictions

b)

Sharing data sources and methods with stakeholders

c)

Encrypting all outputs

d)

Hiding how decisions are made

55.

Which Azure features help support responsible AI?

a)

Responsible AI Dashboard in Azure ML

b)

Content filtering in Azure AI Content Safety

c)

Azure DevOps Pipelines

d)

Azure Monitor

56.

Which actions demonstrate accountability in AI?

a)

A. Assigning clear human ownership for AI decisions

b)

B. Testing and auditing models regularly

c)

C. Removing all human review

d)

D. Automating every decision

57.

What is a potential consequence of using non-representative data in AI training?

a)

Reduced need for model evaluation

b)

Faster model deployment

c)

Improved model accuracy for all users

d)

Biased predictions against certain groups

58.

Which Microsoft Azure service can help detect and mitigate harmful outputs from generative AI models?

a)

Azure Virtual Machines

b)

Azure Logic Apps

c)

Azure AI Content Safety

d)

Azure Blob Storage

59.

What does LLM stand for in the context of artificial intelligence?

a)

Large Language Model

b)

Linguistic Learning Machine

c)

Longitudinal Learning Model

d)

Logical Language Matrix

60.

What does 'GPT' stand for in the context of AI?

a)

Generative Pre-trained Transformer

b)

Generative Predictive Text

c)

General Processing Technology

d)

Graphical Processing Tool

61.

What is the first step in training a transformer model?

a)

Attention

b)

Tokenization

c)

Decoding

d)

Embeddings

62.

How does generative AI create new content, such as text or images?

a)

By using random combinations of elements from its training data.

b)

By applying fixed rules programmed by developers.

c)

By learning patterns in the training data and generating outputs based on these patterns.

d)

By mimicking exact examples from its training data without any variation.

63.

What is the potential impact of foundation models in various industries?

a)

Cause confusion and misinformation in industries

b)

Only be used for academic research

c)

Have no impact on any industry

d)

Revolutionize industries such as healthcare, finance, and customer service

64.

What is the potential problem of hallucinations in generative AI?

a)

They improve the model's accuracy

b)

They have no impact on the model's output

c)

They are necessary for the model to function properly

d)

They can make the output text difficult to understand and generate incorrect information