Unit 1 AI Reflection

Unit 1.1: Understanding AI

• Definition of Artificial Intelligence: AI is the ability of machines to mimic human traits,

such as decision-making, predicting the future, learning, and self-improvement. A
machine is considered intelligent when it can perform tasks independently by collecting,
understanding, analyzing data, learning from it, and improving over time.
• Core Concepts of AI:
o AI is a form of intelligence.
o AI involves both technology and academic study.
o AI systems can perform tasks that typically require human intelligence.
• Applications and Domains of AI: AI impacts various fields and applications, including
but not limited to data analysis, natural language processing (NLP), and computer vision
(CV).
• Goals of AI Curriculum: The program aims to develop a deep understanding of AI,
provide access to AI-powered tools, and enable participants to create AI-based solutions.

Learning Activities and Games:

1. Game 1: Rock, Paper, Scissors (Based on Data for AI)

o Objective: Observe how AI predicts your next move by learning from your
patterns.
o URL: next.rockpaperscissors.ai
2. Game 2: Semantris (Based on Natural Language Processing - NLP)
o Objective: Understand AI’s capability to associate words and understand
language context.
o URL: Semantris
3. Game 3: Quick, Draw (Based on Computer Vision - CV)
o Objective: See how AI interprets and guesses drawings made by humans.
o URL: Quick, Draw

Applications of AI:

• Face Recognition: Used for smartphone face locks.

• Smart Assistants: Siri, Alexa, and others recognize speech patterns and respond
accordingly.
• Fraud Detection: Used in finance to analyze customer data and predict risks.
 • Medical Imaging: Helps create and analyze 3D models from 2D scans for better
diagnosis.

Understanding AI Domains:

• Natural Language Processing (NLP): Deals with textual data, enabling machines to
understand and interact using human language.
• Computer Vision (CV): Focuses on interpreting and processing visual data from images
and videos.
• Data for AI: Utilizes statistical data analysis to draw insights and make decisions.

Reflection and Analysis:

• These games provide hands-on experience in understanding the principles of AI, showing
its application in real-world scenarios.
• AI integrates multiple domains to achieve complex tasks that simulate human
intelligence.

Quiz and Activities for Revision:

• Quiz questions covering AI applications, languages, domains, and their practical uses.
• Activities include teamwork, brainstorming applications of AI, and exploring how AI can
solve current societal issues, like mask detection during pandemics.

Purpose of AI Activities:

• To make students aware of the various domains of AI (NLP, CV, and Data) and
understand their practical implications.
• Encouraging collaborative learning through interactive games and challenges.
Lesson Title: AI Project Cycle

Approach: Interactive Session

Summary: Students will learn about the AI Project Cycle, familiarize themselves with its stages,
and understand how to approach starting an AI project.
Learning Objectives: Students will know how to initiate and develop an AI project effectively.
Learning Outcomes: Students will be able to describe the stages in the AI project cycle and
understand the significance of each stage.
Pre-requisites: Basic computer literacy
Key Concepts: AI project cycle

Let Us Think!

Problem Scoping means

Problem Scoping is identifying and clearly defining a problem that needs to be solved. It
involves understanding the problem's context, stakeholders affected by the problem, and the
ultimate goals of solving it.

Data Acquisition means

Data Acquisition involves gathering the necessary data required for solving the problem. This
data can come from various sources such as databases, sensors, surveys, or online repositories.

Data Exploration means

Data Exploration is the process of analyzing the acquired data to understand its characteristics,
patterns, and relationships. This stage helps in cleaning, organizing, and preparing the data for
further analysis.

Modeling means
Modeling refers to selecting and applying appropriate algorithms and techniques to develop
predictive or analytical models based on the explored data. It involves training the AI to
recognize patterns and make decisions.

Evaluation means
Evaluation is the stage where the developed models are tested and assessed for their accuracy,
reliability, and effectiveness. This involves comparing the model's predictions or outputs against
actual data to measure performance.

Deployment means
Deployment involves implementing the developed AI model into a real-world application,
making it accessible for end-users. It means putting the solution into practice to address the
identified problem.

Let Us Understand!

Example: AI in Pest Management

Problem: Pest infestation damages crops.

Context:

• The cotton industry in India is crucial, with 6 million local farmers relying on it.
• Cotton crops are frequently infected with Pink Bollworm, which is difficult to detect with
the naked eye.
• Small farmers lack advanced tools to protect their crops effectively.

Can AI help?
AI can potentially be used to detect pest infestations early, allowing farmers to take timely
action. An AI system could analyze images of crops to detect pests, recommend appropriate
pesticides, and guide farmers on the correct application timing.

Task:

1. Identify all factors to save cotton crops.

2. Set goals for the AI system, such as early detection of pests and minimizing pesticide
usage.

Data to Collect:

• Images of pests
• Farmer names
• Village names
• Farm sizes
• Pesticide usage records

Data Cleaning:

• Normalize image sizes and formats.

• Remove duplicate data entries.
• Address missing data by filling gaps appropriately.

Data Exploration:

• Identifying patterns in pest infestations.

• Analyzing correlations between pesticide usage and pest infestation.
• Understanding variations in crop yield.

Modeling:

• Developing an AI model that uses computer vision to recognize pests in images.

• Training the model using datasets of labeled pest images.
• Selecting algorithms best suited for image recognition tasks.
Evaluation:

• Testing the AI model's accuracy by comparing its predictions against actual cases.
• Evaluating different AI models to find the most accurate and efficient one.

Deployment:

• Developing an AI-enabled mobile app for farmers to upload images of pests.

• Providing real-time recommendations based on detected pest levels.

Features of the AI Pest Management System (CottonAce App):

• Detects pests using images taken by farmers.

• Warns of potential infestations.
• Advises on the appropriate amount of pesticide.
• Guides on the timing of pesticide application.
• Provides access to professional help if necessary.

Why Use an AI Project Cycle?

1. Efficiency: Breaking down the project into stages ensures that each aspect is thoroughly
considered and executed, leading to faster and more effective solutions.
2. Modularity: Complex problems can be broken into manageable parts, making it easier to
develop and implement solutions.

AI Project Cycle Defined

The AI Project Cycle is a structured, cyclical process that guides the development of AI projects.
It includes multiple stages:

1. Problem Scoping: Defining the problem and setting the project's goals.
2. Data Acquisition: Gathering relevant data to address the problem.
3. Data Exploration: Analyzing the data to uncover insights.
4. Modeling: Building and training AI models using the data.
5. Evaluation: Testing the models to ensure accuracy and reliability.
6. Deployment: Implementing the solution in a real-world environment.

1.2.1 Problem Scoping

Approach: Instructor-led Interactive Session + Activity

Summary: Students will learn about the importance of problem scoping and will use templates
to set goals for their AI projects.
Learning Objectives:

• Understand the importance of problem scoping.

• Learn to use frameworks and templates for defining project goals.

Learning Outcomes:

• Apply the problem scoping framework effectively.

• Frame clear goals for AI projects.

Key Concepts: Problem scoping, 4Ws Problem Canvas

Problem Scoping: Activity and Framework

Objective: Identify a real-world problem that can be addressed using AI.

Steps:

1. Select a theme of interest (e.g., agriculture, healthcare, digital literacy).

2.
3. Narrow down to specific topics under the theme (e.g., pest management in agriculture).
 4. Identify problems within the selected topic.

5.
6. Define the goal of the project using a structured template.

4Ws Problem Canvas:

• Who: Identify stakeholders affected by the problem (e.g., farmers).

• What: Define the nature of the problem (e.g., crop damage due to pests).
• Where: Context and location of the problem (e.g., cotton farms in India).
• Why: Benefits of solving the problem (e.g., increased crop yields, reduced pesticide
costs).

•
Example Problem Statement:
Farmers are experiencing significant crop losses due to undetected pest infestations. An AI-
enabled mobile app could help by providing early detection of pests through image recognition,
advising on pesticide use, and optimizing crop management practices.

Questions for Reflection

1. What are the various stages of the AI Project Cycle? Can you explain each with an
example?
2. How is an AI project different from an IT project?
3. Explain the 4Ws problem canvas in problem scoping.
4. Why is there a need to use a Problem Statement Template during problem scoping?
5. What is Problem Scoping? What are the steps of Problem Scoping?
6. Who are the stakeholders in the problem scoping stage?

Conclusion

The AI Project Cycle provides a systematic approach to solving complex problems using AI. It
allows for greater efficiency and effectiveness by breaking down tasks into manageable stages,
ensuring that all aspects of the project are thoroughly considered and addressed.

Data Acquisition

• Approach: Interactive Session + System Maps

 • Summary: Students will learn how to acquire data from reliable and authentic sources.
They will understand how to analyze data features that affect their problem scope and
learn about System Maps.

Learning Objectives:

1. Students will learn various ways to acquire data.

2. Students will learn about data features.
3. Students will learn about System Maps.

Learning Outcomes:

1. Students will be able to identify the data required for a given problem.
2. Students will be able to draw System Maps.

Pre-requisites:

• Basic computer literacy

Key Concepts:

1. Understanding reliable and authentic data sources.

2. Introduction to System Mapping.

Overview of the Lesson:

• Previous Module Recap:

o Scoping a problem and setting a project goal using the 4Ws problem canvas:

o
1. Who? Identifying stakeholders and what is known about them.
2. What? Understanding the problem and providing evidence.
3. Where? Analyzing the context and location of the problem.
4. Why? Determining value for stakeholders and how solutions can improve
their situation.
 o Crafting a problem statement using identified details.
• Introduction to Data Acquisition:
o Data Acquisition is the process of gathering information or facts to support AI
projects.
o Example: Using past salary data to predict future salaries—training data vs.
testing data.
o Importance: Ensures the AI model is trained with relevant and authentic data for
accurate predictions.
• Concept of Data Features:
o Data features are specific types of data needed to address the problem.
o Example: For salary prediction, features might include salary amount, increment
percentage, increment period, bonuses, etc.
• Acquiring Data from Reliable Sources:
o Emphasis on acquiring data that is reliable, authentic, and relevant.
o Sources of data should be checked for authenticity to avoid incorrect training of
AI models.
o Use government portals for open-sourced data:
▪ Examples: data.gov.in, india.gov.in

Methods of Data Collection:

1. Surveys
2. Web Scraping
3. Sensors
4. Cameras
5. Observations
6. APIs (Application Program Interface)

7.

Introduction to System Maps:

• Purpose: Understand relationships among different elements of a problem and strategize

solutions.
•
• Example: Water Cycle System Map showing relationships between clouds, snow, rivers,
etc.
• Elements of a System Map:
o Arrows show cause and effect.
o Signs (+ or -) indicate direct or inverse relationships.
• Activity: Students will create their own System Map based on identified data features.
 •

Session Preparation:

• Logistics: Group activity (groups of 4) for a class of 40 students.

• Materials Required:
o Computers, chart paper, sketch pens.

Resources:

• Link to online System Map tool: System Map Tool

Revision Questions:

1. Differentiate between Training Data and Testing Data with examples.

2. List methods for data collection and give examples of projects using each method.
3. What should be kept in mind to ensure useful data collection?
4. Draw a system map for a scenario enabling farmers to take produce to market.
5. Name some government websites for acquiring open-source data.

Data Exploration

Approach:

• Activity-Based Learning
Summary:

• Students will explore different types of graphs used in data visualization. They will learn
to find trends and patterns within datasets.

Learning Objectives:

• Explore various types of graphical representations.

• Learn how to visualize the data they have.

Learning Outcomes:

• Recognize different types of graphs used in data visualization.

• Identify patterns and trends from explored data.

Pre-requisites:

• Basic computer literacy.

Key Concepts:

• Data Visualization.

Let us Recap!

• Quiz Questions:
1. What is the second stage of the AI project cycle? (Answer: a. Data Exploration)
2. What comes under Problem Scoping? (Answer: b. 4Ws Canvas)
3. Which is not valid for Data Acquisition? (Answer: d. Announcements)
4. What does an arrow with a minus sign between X and Y mean? (Answer: a. If X
increases, Y decreases)
5. Which is not part of the 4Ws Problem Canvas? (Answer: d. Which?)

Session Preparation

Logistics:

• Class Size: 40 students

• Group Activity: Groups of 4

Materials Required:
 • Computers (10)

Resources:

• Website for visualization techniques: Data Visualization Catalogue

Purpose:

• Understand the importance of data exploration before training AI models.

Instructional Approach:

• Relate data exploration to skimming a book before reading, to understand its suitability.
• Visualization helps to quickly comprehend trends, relationships, and patterns, guiding
model selection.

Activity Brief:

• Task: Explore various types of graphs using an online resource. Learn new visualization
techniques.
• Intervention: Guide students to choose suitable graphs for their specific data features.

Steps:

1. Visit the visualization website to explore different graphical representations.

2. Identify five new data visualization techniques and document:
o Name of Representation
o Description
o How to draw it
o Suitable data types

Sketchy Graphs Activity:

Purpose:

• Learn different visualization techniques and use the appropriate graph to represent data.

Instructions:

1. Students sketch graphs based on the data features they have identified.
2. Select two data features, plot their graphs on chart paper, and relate them to the project
goal.
 3. Present the graphs, describing observed trends or patterns.

Materials Required:

• Chart Paper (10), Sketch-pens (10), Ruler (10), Basic Stationary (10 sets)

Reflection and Analysis:

• Post-Activity Reflection:
o What data feature are you representing?
o Why did you choose this representation?
o List down trends or patterns observed.

Revision Questions:

1. Why is Data Exploration significant after data acquisition?

2. What is the relevance of Data Visualization in AI?
3. List five graphs used for data visualization.
4. How does Data Exploration differ from Data Acquisition?
5. Provide an example to explain a Data Visualization technique.

These notes capture the essence of the "Data Exploration" activity and guide students through the
process of understanding and applying data visualization techniques.

Different visualisation Methods

Notes on AI, ML, and DL

Overview

• Previous Module: Data Exploration involved analyzing data acquired during the Data Acquisition
stage.
• Current Objective: Using insights from data exploration to develop a strategy for AI-based
projects.
• AI Project Development: Involves building or utilizing AI models/algorithms.
• Clarification: Understanding the distinctions between Artificial Intelligence (AI), Machine
Learning (ML), and Deep Learning (DL).

AI, ML, & DL Definitions

1. Artificial Intelligence (AI):

o Enables computers to mimic human intelligence.
o Works on algorithms and data to provide desired outcomes.
2. Machine Learning (ML):
o A subset of AI.
o Allows machines to improve at tasks with experience.
o Learns from new data and adjusts based on past mistakes.
3. Deep Learning (DL):
o A subset of ML.
o Involves training models on vast datasets.
o Uses multiple ML algorithms to improve task performance.

o
Relationship Between AI, ML, and DL

• AI: The broad umbrella term covering all intelligent machine systems.
• ML: A subset of AI, focusing on learning from data.
• DL: A more specific form of ML, dealing with large-scale data and complex models.

Modelling Approaches

• Rule-Based Approach:
o The machine operates based on predefined rules.
o Example: A dataset predicting whether a child can play golf based on weather
conditions.

o
• Learning-Based Approach:
o The machine learns from data without predefined rules.
o Adapts to changes and improves over time.
o Example: Distinguishing between images of apples and bananas by training on labeled
images.
 o

Purpose of AI Modelling

• Developing algorithms (models) to generate intelligent outputs.

• Involves a mathematical approach to analyzing data relationships.

Questions for Reflection

1. What makes a machine intelligent?

2. How can a machine become artificially intelligent?
3. Can AI be a threat to human intelligence? If so, how?

AI Project Cycle Stages

1. Problem Scoping: Identifying the problem and defining objectives.

 2. Data Acquisition: Gathering relevant data for analysis.
3. Data Exploration: Analyzing data to identify patterns and trends.
4. Modelling: Building AI models to make predictions or decisions.
5. Evaluation: Assessing model performance and making improvements.

Real-Life AI Examples

• Personal assistants (e.g., Siri, Alexa).

• Recommendation systems (e.g., Netflix, Amazon).

Comparing Rule-Based and Learning-Based Models

• Rule-Based:
o Static learning.
o Suitable for situations with clear, predefined rules.
• Learning-Based:
o Dynamic learning and adaptation.
o Suitable for complex scenarios with varying data patterns.

Classification of AI Tasks

1. Classification: Making a diagnosis based on symptoms.

2. Regression: Predicting house prices.
3. Clustering: Shortlisting job applications.
4. Classification: Detecting credit card fraud.
5. Classification: Filtering spam emails.

These notes summarize the essential concepts regarding AI, ML, and DL, highlighting their
definitions, interrelationships, and approaches to modeling, which are foundational for building
AI-driven solutions.

AI Model Evaluation

Overview

• Stage 5 of the AI project cycle focuses on Evaluation.

• The main goal of this stage is to test different models and select the best one.

Lesson Title: Evaluation

• Approach: Interactive session combined with an activity.

• Summary: This module teaches the concept of evaluation in the AI project cycle. Evaluation is
essential for:
o Assessing the success of AI projects.
o Identifying areas for improvement.
o Making data-driven decisions.
Learning Objectives

1. Understand the importance of evaluation in the AI project cycle.

2. Apply evaluation techniques to assess the effectiveness of AI projects.
3. Identify areas for improvement in AI projects through evaluation.

Learning Outcomes

• By the end of the lesson, students should be able to apply evaluation techniques in their AI
projects.

Pre-requisites

• Basic knowledge of Artificial Intelligence and problem-solving skills.

Key Concepts

1. Importance of Evaluation: It ensures reliability, helps improve models, and confirms that the AI
solution is solving the intended problem effectively.
2. Definition of Evaluation:
o The process of understanding the reliability and performance of an AI model.
o Involves feeding a test dataset into the model and comparing outputs with actual
answers.
o Avoid using training data for evaluation to prevent overfitting (where the model simply
memorizes the data).

Evaluation Process

• Testing Models:
•
o Use testing data to check model performance.
o Evaluate how well the model addresses the AI problem.
o Performance is calculated based on specific parameters.

Model Evaluation Techniques

• Evaluation involves different techniques depending on the model type and purpose.
• Example: Testing a forest fire prediction model:
o True Positive (TP): The model correctly predicts a fire when there is one.
o
o False Positive (FP): The model incorrectly predicts a fire when there isn't one.
o
o False Negative (FN): The model fails to predict a fire when there is one.
o
o True Negative (TN): The model correctly predicts no fire when there isn't one.
 o

Example Scenario: Forest Fire Prediction Model

• True Positive: Model predicts fire, and there is a fire.

• False Positive: Model predicts fire, but there is no fire.
• False Negative: Model predicts no fire, but there is a fire.
• True Negative: Model predicts no fire, and there is no fire.
 •

Use Case: Exoplanet Prediction

• Objective: Evaluate which algorithm makes the best prediction.

• Metrics: Metrics like ROC (Receiver Operating Characteristic) are used to determine model
accuracy.
• Comparison: Models are compared based on their prediction accuracy.

•
Chapter Review Questions

1. What is Evaluation?: It’s the process of testing a model’s performance using test data to check
its reliability and accuracy.
2. What are various model evaluation techniques?: Techniques include checking metrics like True
Positive, False Positive, ROC, and others depending on the model type.
3. Why is model evaluation important in AI projects?: Evaluation helps in verifying the
effectiveness of the model, identifying errors, and ensuring the model is suitable for real-world
application.
4. What do you understand by the terms True Positive and False Positive?
o True Positive (TP): Correct prediction when the actual condition is true.
o False Positive (FP): Incorrect prediction when the actual condition is false.

These notes highlight the significance of model evaluation in AI projects, focusing on assessing
model reliability, understanding various evaluation terminologies, and emphasizing the need for
data-driven decision-making.

Lesson Title: Deployment

Approach: Interactive Session + Activity

Summary:

• In this module, students will learn about the term "deployment" in the context of AI projects
and understand why it is a crucial step.
• Students will connect the concept of deployment to real-world examples, such as deploying a
chatbot on a website or a predictive model in a mobile app.

Learning Objectives:

• Understand the concept of deployment within the AI project cycle.

• Demonstrate knowledge through hands-on activities related to deployment.

Learning Outcomes:

• Emphasize the importance of deployment in the AI project cycle.

• Apply knowledge of deployment in future AI projects.
• Encourage exploration of different deployment methods.

Pre-requisites:

• Basic knowledge of Artificial Intelligence and problem-solving skills.

Key Concepts:

1. Importance of Deployment in AI Project Cycle:

 o Deployment is the final stage in the AI project cycle where the AI model or solution is
implemented in a real-world scenario. It is critical because it makes the solution
available to users, allowing it to generate value.
2. What is Deployment?
o Deployment involves the integration of the AI model into a production environment
where it can interact with users and other systems.
3. Key Steps in the Deployment Process:
o Testing and Validation: Ensuring that the AI model performs accurately and effectively.
o Integration: Incorporating the model into existing systems, such as software
applications or websites.
o Monitoring and Maintenance: Regularly checking the model’s performance and making
necessary updates or improvements.
4. Examples of Successful AI Project Deployments:
o Self-driving cars
o Medical diagnosis systems
o Chatbots used for customer service
5. Deployment Platforms:
o Mobile Applications: AI models can be deployed on mobile apps to provide real-time
predictions and insights.
o Website Applications: AI models can be integrated into web applications to enhance
user interactions and services.

Activity Time!

Activity Title: Implementation of AI Project Cycle to Develop an AI Model for Personalized

Education

• Activity Introduction:
In this activity, students will use the AI project cycle to conceptualize a solution for the
given problem.
o The AI project cycle is a 6-step process that helps in problem-solving using Artificial
Intelligence.
• Description:
All individuals have different cognitive levels and personalities, necessitating a
personalized education approach. A generalized education system does not cater to these
individual needs.
• Activity Guidelines:
o Understand the problem: Personalized education for each student.
o Learn the various aspects and developments in the field of AI in education.
o Fill in the AI Project Cycle mapping template for the problem.
o The solution will involve an AI algorithm that learns from student behavior and
preferences to provide tailored educational content.

Case Study: Preventable Blindness

Problem:
 • Preventing vision loss and reducing delays in report generation for diabetic retinopathy.
• Millions of adults live with diabetes, which can lead to diabetic retinopathy, a condition that can
cause blindness.

Solution:

• AI models are used to detect diabetic retinopathy by analyzing images of the eye.
• At Aravind Eye Hospital in India, an AI screening solution, developed with Google, has achieved
high accuracy in detecting the condition.
• This AI solution allows for quick detection, even in rural areas, without the need for specialized
doctors.

AI Project Cycle Mapping Template for Preventable Blindness:

 1. Problem Scoping:
o Defining the problem of preventing vision loss due to diabetic retinopathy.
2. Data Acquisition:
o Collecting retinal images from patients in clinics using cameras.
3. Data Exploration:
o Validating and analyzing the collected data to create a useful dataset.
4. Modeling:
o Developing an AI model that can diagnose diabetic retinopathy based on the retinal
images.
5. Evaluation:
o Testing the model for accuracy and refining it to achieve the desired output.
6. Deployment:
o Implementing the AI model in tools that can be used in clinics, especially in remote
areas.

Revision Time

1. Choose the Correct Answer:

1. Does modeling mean creating an AI model?
a. YES
b. NO
2. Can we use AI on mobile phones?
a. YES
b. NO
3. What is deployment in the context of an AI project cycle?
4. Why is deployment an important phase in the AI project cycle?
5. What are some common challenges in deploying AI models?
2. Rearrange the Steps of the AI Project Cycle in the Correct Order:
o a. Data Acquisition
o b. Problem Scoping
o c. Modeling
o d. Data Exploration
 o e. Deployment
o f. Evaluation
3. Identify A, B, and C in the Following Diagram (Hint: How AI, ML & DL are
Related to Each Other).

Conclusion

• Emphasize the role of deployment as a crucial final stage in the AI project cycle.
• Encourage students to think about how they can use AI deployment techniques in future
projects and the challenges they might face.

Title: AI Ethical Issues

Approach: Interactive Session + Activity

Summary:

Students will learn about the differences between morals and ethics, understand ethical issues
related to personal data, and learn ethical principles for creating safer AI systems.

Objectives:

1. Understand the concept of ethics and morals.

2. Learn to differentiate between morality and ethics.
3. Explore ethical considerations related to personal data, AI ethics, and AI ethics principles.

Pre-requisites:

1. Basic knowledge of the AI project cycle and its steps.

2. Basic understanding of ethics and ethics in AI.

Key Concepts:

1. Familiarization with the AI project cycle, its need, and how it maps to different projects.
2. Understanding AI ethics and the issues around AI ethics.
3. Ethical principles for creating safer AI solutions.

Ethical Scenarios for Discussion

Ethical Scenario I

Imagine you're a high school teacher with many essays to grade. You discover an AI tool that
can automatically grade essays and assign them grades.
Discussion Questions:

1. Would you use the AI tool to grade the essays? Why or why not?
2. What are the advantages and disadvantages of using the AI tool?
3. Can you think of any challenges or potential biases the AI tool might face?

Ethical Scenario II

You're overseeing a busy fast-food restaurant. During a rush, you drop a burger on the dirty
floor. Your boss sees this and tells you to pick it up and serve it.

Discussion Questions:

1. Would you serve the burger? Why or why not?

2. What ethical considerations are involved in this scenario?
3. How would your decision impact customers, your boss, and yourself?

Examples of Ethical Questions

1. Is it ethical to keep extra change accidentally given to you by a shopkeeper?

2. Is taking pens from a library considered stealing?
3. If you buy a dress, wear it to an event, and then return it claiming it doesn't fit, is that ethical?

Examples of Moral Questions

1. Is it ever acceptable to lie? If so, under what circumstances?

2. If a family is starving, is it morally acceptable for them to steal food from a wealthy store owner?
3. Is a decision made collectively by a group always morally right?

Understanding Ethics vs. Morals

Morals

• The beliefs dictated by society, culture, or religion.

• Examples: Always speak the truth, always be loyal.

Ethics

• Guiding principles that help decide what is right or wrong.

• Examples: Is it ethical to always tell the truth even if it might harm someone?

Key Difference:

• Morals are subjective and can vary based on cultural or societal norms.
 • Ethics are more universal principles of right and wrong, often used to guide professional
behavior.

Fun Activity: Moral Machine

Purpose:

• Use the Moral Machine platform to explore moral decisions made by AI, such as in self-driving
cars.

Guidelines:

1. Go to the Moral Machine.

2. Click on "Start Judging" to begin the activity.
3. Answer the scenarios until the end.
4. Discuss the results, which reflect how personal beliefs influence moral decisions.

Discussion Points:

1. How does your decision-making align with others?

2. What do your choices reveal about your ethical and moral values?
3. How would these decisions impact AI programming in real-world situations?

Ethics and Personal Data

Scenario: Jack, a student, spends a lot of time online for research, social media, gaming, and
shopping. This means a lot of his personal information is stored on the internet.

Discussion Questions:
 1. What types of personal data might be stored online?
2. How could this data be used to influence Jack’s decisions?
3. Is it ethical for governments or companies to access personal data without consent?

Major Issues around AI Ethics: Example of Jack

Jack is concerned about an AI-based essay grading system at his school that gave him a bad
grade, even though he wrote a good essay.

• Possible reasons for the error:

o The AI was trained on biased data, possibly from a different country or demographic.
o The training data did not represent Jack’s context, leading to incorrect evaluations.

Principles of AI Ethics
1. Human Rights:
o AI should respect and protect human rights.
o AI must not take away freedom, discriminate, or deprive people of jobs.

o
2. Bias:
o AI should avoid biases stemming from skewed training data.
o AI must be trained on diverse datasets to represent all populations fairly.

o
3. Privacy:
o AI must ensure data privacy and safety.
o AI should be transparent about the data it collects and its purpose.
 o
4. Inclusion:
o AI must be inclusive, ensuring it benefits everyone equally.
o AI solutions should be accessible and usable by diverse groups.

Revision and Key Takeaways

1. Ethics are guiding principles that help decide what is good or bad.
2. AI solutions should follow ethical guidelines to ensure safety, fairness, and inclusivity.
3. Moral Questions: What society considers right or wrong.
4. Ethical Questions: Whether actions are justifiable based on universal principles.

Questions to Assess Understanding:

1. Differentiate between ethics and morals with examples.

2. Define principles of AI ethics.
3. Discuss the importance of data privacy in AI.
4. How should inclusivity be addressed when developing AI models?
5. Identify and explain major ethical issues around AI, like bias and privacy.
Concluding Remarks

Understanding ethics and morality is essential for developing AI that is fair, inclusive, and
respects human rights. Students must learn to critically evaluate the impact of AI on society and
adhere to ethical principles in its development and deployment.

Unit 2 - Data Literacy

Unit 2.1 - Basics of Data Literacy

Lesson Title: Basics of Data Literacy

Approach: Session + Activity
Summary: This module introduces students to the concept of Data Literacy, its importance, and
its impact on decision-making and daily life. Students will learn about the different categories of
data and gain an understanding of data privacy and security.

Learning Objectives:

• Define data literacy and explain its significance using real-world examples.
• Understand the impact of data literacy on daily life.
• Develop awareness of personal data, data privacy, and data security.

Learning Outcomes:

• Define data literacy and recognize its importance.

• Understand how data literacy supports informed decision-making and critical thinking.
• Apply the Data Literacy Process Framework to analyze and interpret data.
• Differentiate between data privacy and security.
• Identify risks associated with data breaches and unauthorized access.
• Learn measures to protect data privacy and enhance data security.

Pre-requisites: Basic knowledge of AI and data

Key Concepts:

• Understanding data literacy

• Difference between Quantitative (Numerical) and Qualitative (Categorical) Data
• Impact of data literacy through case studies and scenarios
• Best practices for Cyber Security

2.1.1 Introduction to Data Literacy

Data literacy involves understanding, working with, and communicating data. It encompasses the
ability to collect, analyze, and present data meaningfully.
Data Pyramid Stages:

1. Raw Data: Unprocessed facts or information.

2. Information: Processed data providing insight about the world.
3. Knowledge: Understanding how things are happening.
4. Wisdom: Understanding why things are happening in a particular way.

Example: Rahul rated three films as bad, best, and average. The data is qualitative and
categorical. Filtering this data involves interpreting subjective opinions into categories.

Data Pyramid with a simple Traffic Light example:

Reference Video: Introduction to Data Literacy

2.1.2 Impact of Data Literacy

Activity: Impact of News Articles

Session Preparation Logistics: For a class of 40 students [Pair Activity]
Materials Required: Computers (20), Online Data Sources

Task: Rank news sources from most accurate to least, providing reasons for your choices.
Purpose: This activity aims to help participants practice evaluating and validating data sources
to develop critical data literacy skills.

Purpose of the Activity: To engage participants in understanding the importance of reliable data
sources and the impact of data literacy on decision-making.

2.1.3 How to Become Data Literate

Example Scenario: Buying a video game online

A data literate person can:

• Filter products based on price and requirements.

• Check user ratings and reviews.
• Ensure the product meets specific needs.

Data Literacy Process Framework: An iterative process guiding effective data use and
awareness.
2.1.4 What are Data Security and Privacy? How are They Related to AI?

Data Privacy:

• Concerned with proper handling and protection of sensitive and personal data.
• Examples: Data breaches at government agencies, hospitals, or corporations.

Best Practices for Data Privacy:

• Understand data handling and storage.

• Collect only necessary data.
• Obtain user consent during data collection.
Data Security:

• Protects digital information from unauthorized access, corruption, or theft.

Importance of Data Security:

• Rising cyber threats and technological advancements make data security crucial.

2.1.5 Best Practices for Cyber Security

Do’s:

• Use strong, unique passwords and Two-Factor Authentication (2FA).

• Download from trusted sources and scan files.
• Use secure Wi-Fi and update software regularly.
• Adjust social media privacy settings and lock screens.

Don’ts:

• Avoid sharing personal info, opening suspicious emails, and engaging in cyberbullying.

Reference Links:

• Cyber Security Video

• CBSE Manual on Cyber Security

Revision Questions:

1. Cultivating Data Literacy means:

o b) Acquire, develop, and improve data literacy skills
2. Data Privacy and Data Security are often used interchangeably but they are different from
each other:
o a) True
3. The _______________ provides guidance on using data efficiently and with all levels of
awareness:
o b) Data literacy framework
4. _____________ allows us to understand why things are happening in a particular way:
o d) Wisdom
5. ____________ is the practice of protecting digital information from unauthorized access,
corruption, or theft throughout its entire lifecycle:
o a) Data security

Unit 2.2 - Acquiring Data, Processing, and Interpreting Data

Lesson Title: Acquiring Data, Processing, and Interpreting Data

Approach: Session + Activity
Summary: This lesson focuses on data acquisition, processing, and interpretation, introducing
key terminologies and methods.

Learning Objectives:

• Familiarize students with data terminologies and methods.

• Discuss data interpretation methods and their advantages/disadvantages.
• Recognize various data presentation methods and their impact on business.

Learning Outcomes:

• Determine the best methods for acquiring data.

• Classify different types of data and acquisition methodologies.
• Define and describe data interpretation methods.
• Realize the importance of data interpretation.

Pre-requisites: Acquaintance with data and its different types.

Key Concepts:

• Data terminologies (acquisition, processing, analysis, presentation, interpretation)

• Quantitative and Qualitative Data Interpretation
• Types of Data Interpretation (Textual, Tabular, Graphical)

2.2.1 Types of Data

Textual Data (Qualitative):

• Made up of words and phrases.

• Used for Natural Language Processing (NLP).
• Example: Search queries on the internet.

Numeric Data (Quantitative):

• Made up of numbers.
• Used for statistical analysis.
• Example: Cricket scores, restaurant bills.

Numeric Data Classification:

• Continuous Data: Height, weight, temperature.

• Discrete Data: Number of students in a class.

2.2.2 Data Acquisition

Steps in Data Acquisition:

Acquiring Data – Sample Data Discovery
Let’s say we want to collect data for making a CV model for a self-driving car

We will require pictures of roads and the objects on roads

We can search and download this data from the internet
This process is called data discovery

Acquiring Data – Sample Data Augmentation

Data augmentation means increasing the amount of data by adding copies of existing data with

small changes
The image given here does not change, but we get data on the image by changing different
parameters like color and brightness
New data is added by slightly changing the existing data

Acquiring Data – Sample Data Generation

Data generation refers to generating or recording data using sensors
Recording temperature readings of a building is an example of data generation
Recorded data is stored in a computer in a suitable form

1. Data Discovery: Searching for relevant datasets (e.g., images for a self-driving car model).
2. Data Augmentation: Increasing data by making slight changes (e.g., altering image color and
brightness).
3. Data Generation: Using sensors to record data (e.g., temperature readings).

Sources of Data:

• Primary Data: Surveys, interviews, experiments.

•
• Secondary Data: External sources.
 •

2.2.3 Best Practices for Acquiring Data

Checklist:

• Ensure data is relevant and high-quality.

• Address ethical concerns in data acquisition.

Ethical concerns in data acquisition

• While gathering data and choosing datasets, certain ethical issues can be addressed before
they occur
 •

2.2.4 Features of Data and Data Preprocessing

Usability of Data:

1. Structure: How data is stored.

2.
3. Cleanliness: Data should be free from duplicates and anomalies.
 4.
5. Accuracy: Data should match real-world values.

6.

Features of Data:
• Independent Features: Input for predictions.
• Dependent Features: Output of predictions.

•
Questions for Consideration:

• What makes data more usable?

• Is unclean data good for AI?

Data Processing and Data Interpretation

Overview

Data processing and interpretation have become essential in today’s data-driven world, enabling
us to derive meaningful insights and make informed decisions.

Introduction to Data Processing

• Definition: Data processing involves using computers to transform raw data into a readable and
interpretable format. It is essential for making sense of the data collected.
• Purpose: Helps in converting unorganized raw data into structured information that can be
utilized for various operations.
• Example: Calculating the total number of candies owned by Niki (7) and Ruchi (4) can be done
through data processing.

Data Interpretation

• Definition: Data interpretation is the process of making sense of processed data. It allows us to
answer specific questions by deriving meaning from the data.
• Purpose: Enables the understanding of trends, patterns, and answers to critical questions.
• Example: Determining how many more candies Ruchi needs to have an equal number with Niki
is an example of data interpretation.

Understanding Key Terms

1. Acquire Data: Collecting data from various sources.

2. Data Processing: Transforming raw data into useful information.
 3. Data Analysis: Examining data components to draw conclusions.
4. Data Interpretation: Explaining what the analyzed data means in context.
5. Data Presentation: Organizing and presenting data in a logical format.

6.

Types of Data Interpretation

1. Quantitative Data Interpretation:

o Focuses on numerical data.
o Answers questions such as "how many," "how often," and "when."
o Examples: Number of website visits, number of likes on an Instagram post.
 o
o
2. Qualitative Data Interpretation:
o Focuses on non-numerical data such as emotions and motivations.
o Methods include observations, focus groups, case studies, and one-to-one interviews.
o Examples: Customer reviews, feelings about a product.

Methods of Data Interpretation

• Qualitative Data Methods: Record keeping, observation, case studies, focus groups, longitudinal
studies, and interviews.
• Quantitative Data Methods: Surveys, polls, observations, and longitudinal studies.
 •

Steps in Data Interpretation

1. Collect Data: Gather information from reliable sources.

2. Organize Data: Sort and arrange the data systematically.
3. Analyze Data: Use statistical tools to evaluate the data.
4. Interpret Data: Derive meaning and draw conclusions.
5. Present Data: Use graphs, charts, and reports to present findings.

Types of Data Presentation

1. Textual Data Interpretation (DI): Presenting data in written form. Suitable for small datasets.

2.
 3. Tabular DI: Representing data using tables with rows and columns. Suitable for structured data.

4.
5. Graphical DI: Using visual elements like bar graphs, pie charts, and line graphs for better data
comprehension.
6. a Bar Graph, data is represented using vertical and horizontal bars.
7.

8.

Pie Charts
Pie Charts have the shape of a pie and each slice of the pie represents the portion of the entire pie
allocated to each category
It is a circular chart divided into various sections (think of a cake cut into slices)
Each section of the pie chart is proportional to the corresponding value
Line Graphs
A line graph is created by connecting various data points.
It shows the change in quantity over time.

Example Activities for Data Interpretation

1. Trend Analysis: Using tools like Google Trends to analyze trending topics over time.
2. Visualizing Student Performance: Creating bar graphs and pie charts to display student scores
across different subjects.
Importance of Data Interpretation

• Informed Decision Making: Provides a factual basis for decisions.

• Cost Reduction: Helps identify areas for improvement or optimization.
• Identifying Needs: Understand consumer preferences or requirements.

Summary: Qualitative vs. Quantitative Data Interpretation

• Qualitative Data Interpretation: Focus on emotions and motivations. Methods include

interviews and focus groups.
• Quantitative Data Interpretation: Numerical and statistical analysis. Methods include polls and
surveys.

Activities for Practical Understanding

 • Visualize and Interpret Data: Engage with tools like Tableau for data visualization.
• Activity Using Tableau: Collect and visualize data about favorite songs, genres, or other
categories using Tableau software or alternatives like MS Excel and Datawrapper.

These notes outline the fundamental aspects of data processing and interpretation, the methods
used, and the significance of these processes in real-world scenarios.