UNIT-1 : AI REFLECTION, PROJECT CYCLE AND ETHICS

UNIT-2 : DATA LITERACY

UNIT-3 : MATH FOR AI (STATISTICS & PROBABILITY)
UNIT-4 : INTRODUCTION TO GENERATIVE AI
UNIT-5 : INTRODUCTION TO PYTHON
 UNIT AI REFLECTION, PROJECT
1 CYCLE AND ETHICS

SESSION-1.1 AI REFLECTION

LEARNING OUTCOMES
• Identify and appreciate Artificial Intelligence and describe its applications in daily life.
• Recognize, engage and relate with the three realms of AI : Computer Vision, Data Statistics and
Natural Language Processing.

Understanding Artificial Intelligence

The concept of inanimate objects becoming intelligent beings has been around since
ancient civilizations. The Golem, an artificially created human from Jewish tradition, has
long served as a metaphor for Artificial Intelligence (AI), inspiring scientists, philosophers,
writers, and artists.
The term “Artificial Intelligence” itself was coined by John McCarthy in 1956, who is
often referred to as the Father of AI. McCarthy defined AI as "the science and engineering
of making intelligent machines." Since then, AI development has progressed significantly,
leading to its integration into many aspects of our daily lives.
AI refers to the capability of machines to perform cognitive tasks such as thinking, perceiving, learning, problem-
solving, and decision-making, mimicking the way humans use their brains to understand and interact with the world.
When a machine can mimic human traits such as decision-making, predicting the future, learning, and self-
improvement, it is said to possess artificial intelligence. In other words, a machine is considered artificially intelligent
when it can perform tasks independently, including collecting, understanding, analyzing data, learning from it, and
improving its performance.
Artificial Intelligence (AI) is the branch of computer science that involves creating machines
capable of performing tasks that typically require human intelligence, such as learning, reasoning,
problem-solving, and decision-making.
AI works by using large amounts of data and special sets of rules, called algorithms, to learn from the data, recognize
patterns, and make decisions.

DATA
ARTIFICIALLY
+ INTELLIGENT
MACHINE
ALGORITHM

R.K.- Artificial Intelligence - IX Part-B 1

 Did you know?
On May 11, 1997, an IBM computer called IBM Deep Blue beat the World Chess
Champion Garry Kasparov after a six-game match.
Google's Alpha Go and IBM's Deep Blue system designed to play strategic games are
examples of reactive machines.

Write some AI applications that you use in your daily life.

______________________________________________________________________________________________
______________________________________________________________________________________________
______________________________________________________________________________________________
______________________________________________________________________________________________
______________________________________________________________________________________________
______________________________________________________________________________________________

Applications of AI in Everyday Life

Due to its tremendous power, AI has become one of the most
impactful technologies today, influencing every aspect of our lives.
Slowly and steadily, AI is integrating into various facets of daily life,
creating significant changes. We see the use of AI in driverless cars,
personal assistants (like Siri, Alexa, Google Assistant, Cortana), smart
speakers (such as Google Nest Audio, Apple HomePod Mini, Sonos
One), and services like Netflix, Cogito, drones, chatbots, smart replies,
and search filters in Gmail. Additionally, AI-based cameras are becoming
more prevalent.
AI in Education & Training. As educating 21st century learners is
a challenging task, there is a great scope of AI in this field. AI systems are
used for predicting performance, designing curriculum, smart
assessments, assisting teachers in doing their routine tasks, e-
learning,educational research, automated training systems, virtual
reality-based trainings, 3D learning, robot assisted teaching and training
etc.
AI in Customer Support Systems. AI has revolutionised the
customer suppor t systems and Interactive Voice Response
Systems(IVRS) using smart searches and Natural Language Processing
(NLP)technologies. Chatbot enabled chat systems used now a days in
customercare services are based on NLP. Chatbots, not only understands text but are also equipped with voice
recognition system to respond to the queries of customers. Google Assistant, Amazon’s Alexa, Apple’s Siri are the
smart assistants which understands human commands and responds to it accordingly. Another example could be AI
enabled alert systems which are used in homes and industries for security. These systems identify an unauthorized
face in the area and raises alarm.
AI in Service Oriented Businesses. In service-oriented businesses like banking, education, travel, them a in
assetis data. AI systems accept the data, works on it, and help its users in bringing out positive results for the business.

2 AI REFLECTION, PROJECT CYCLE AND ETHICS

These systems are also used to understand customer behaviour, customer loyalty, and service quality, which then
helps in predicting future trends and lot more.
AI in Product Oriented Businesses. In product-oriented businesses AI can help in product planning,
designing, manufacturing, and delivering of the products to the customers. AI systems are implemented in various
industries like manufacturing units, construction etc. Autonomous vehicles, smart home devices, smartphones,
robots are some of the potential applications of AI in product-oriented businesses.
AI in e-commerce and Retail Businesses. These days e-commerce is anever-growing industry. Flipkart,
Amazon, Walmart are various popular e-commerce platforms that uses AI to get business and increase customer
experiences. On these platforms AI systems collect bulk amount of data which is used to analyse the popular items,
comparison of items, customer’s experience, feedback, etc.,
AI in Social Media. Social media is another area
where bulk data is generated every moment. This bulk data
can be used for various purposes like promotion of products
and services and enhancing user experiences etc.
Entertainment and media industry are totally dependent on
viewership and subscription by the users. AI system can help
in advanced analysis of viewership data and market trends to
predict future. With the use of intelligent responses, user
privacy and security mechanism, intelligent Al systems can
help to grow business of social media platforms.
AI in Healthcare. AI can be used in healthcare to such
an extent that it can assist in surgery. AI systems can also help
in diagnosis, decision support system in health emergencies,
hospital management and safety, patient rehabilitation,
research, and analysisetc.
AI in Research and Development . AI has
tremendous potential in the research and development in
various fields like health, automobile, population,
environment, defense and many more fields. AI is capable to
process variety of data, that too very fast and are able to
generate patterns for future predictions. This is done using the techniques like NLP, computer vision and deep
learning.

Match the following

SCENARIO APPLICATION
An online retailer uses algorithms to recommend products based on a user’s past Customer Support Systems
purchases and browsing behavior.
A system predicts the next trend in social media based on user interactions and e-Commerce
engagement patterns.
A company deploys an AI chatbot to handle customer queries and provide Social Media
support 24/7.

R.K.- Artificial Intelligence - IX Part-B 3

Domains of AI
The main aim of AI is to develop machines with
human like intelligence. As we know that human
intelligence is made up of domains like vision,
perception, linguistics, learning and reasoning, similarly
AI is composed of the following three domains—
1. Data
2. Computer Vision
3. Natural Language Processing (NLP)
1. Data. Data is considered the key domain out of the three domains of AI. AI system is capable of processing
huge amounts of data at a very high speed and
within fractions of time, and more the data better is
the machine. We can say that data is the
foundation for AI, and all the other capabilities of
AI also revolve around data. Data is basically a
collection of raw facts and figures and can be in
any of the form i.e. simple text, audio visuals, big
data like predictions, insights, forecasts, decisions
etc. And a constant in flow of data is a prerequisite
for an AI system to function. An AI system collects
the data, analyses the data, and then tries to
identify some sort of trend or pattern in it
depending on the requirement. For example, AI
system can be implemented on an online shopping Data Used
portal to predict about customer behaviour which
is based on what a customer buys and how much Numerical Alphanumerical
time a customer spends on a particular part of the
portal. With higher number of customer visits and purchase by the customer the more accurate will be future
prediction done by the AI system.
2. Computer Vision(CV). Computer vision refers
to that domain of AI which enables a machine to
analyse constant feed of huge amount of visual
data, understand various patterns in it and finally
make decisions based on the findings. Now a days,
an enormous amount of data is available in the
form of visuals that is images and videos. The
computer vision can help in doing analysis and
understanding such data by teaching the machines
to see, read and analyse these images and videos
using specialised algorithms.
Data Used
Computer Vision(CV)is used in the following
areas—
Images Videos
(1) Face Recognition—AI systems use face
detection algorithms to recognise faces in the
images.

4 AI REFLECTION, PROJECT CYCLE AND ETHICS

 (2) Image Retrieval—AI systems use content based search for retrieving relevant images.
(3) Smart Cars—in self-driving smart cars computer vision is used to detect traffic signals, signs and lights
for smooth movement of the vehicle.
3. Natural Language Processing (NLP). NLP is a
domain that is broadly defined as the automatic
manipulation of natural language likes speech and
text by a machine. In other words, we can say NLP
enabled interaction between humans and
computer system is done by integrating what has
been said by the user. NLP works using special
algorithms which convert the audio data or speech
into a form that is identified and understood by an
Data Used
AI system. With the help of rules of a language
semantics and pronunciation rules, algorithms try Text Speech / Audio
to interpret what has been spoken and then
digitalises it into a form that is understood by the computer to perform a particular task.
Smart assistants and google translate are some of the AI applications that use NLP. NLP basically has two
main components viz. as follows—
(1) Natural Language Understanding (NLU). NLU involves mapping ofthe input given in a natural
language into useful representations for machine and analysing different aspects of the language.
(2) Natural Language Generation (NLG). NLG is the process of producing meaningful phrases and
sentences generated by the machine in the form of natural language using some machine
representations. NLG involves text planning, sentence planning, and text realisation.

Identify the domain:

A banking system uses AI to analyze transaction data to detect fraudulent activities. It continuously monitors
transactions for patterns indicative of fraud and flags suspicious activity for further investigation.
______________________________________________________________________________________________
A customer service chatbot understands and responds to user queries using natural language. The chatbot can
interpret and process various customer requests, providing accurate and relevant information in real-time.
______________________________________________________________________________________________
A security system uses facial recognition to identify unauthorized individuals entering a restricted area. The system
scans faces, compares them with a database of authorized personnel, and alerts security if an unauthorized person is
detected.
______________________________________________________________________________________________

The three domains of AI—data, computer vision, and natural language processing (NLP)—reveals their close
interconnections. Data serves as the foundation for both computer vision and NLP. Visual data is essential for
computer vision, while audio data is crucial for NLP. When data is presented in audiovisual formats, both computer
vision and NLP collaborate to process and interpret the information.
Communication can be categorized into visual communication and language communication. We
communicate through speech, gestures, signs, signals, expressions, text, and writing, all of which fall under either
visual or language categories. Similarly, AI systems utilize visuals and language for their operations.

R.K.- Artificial Intelligence - IX Part-B 5

 A system receiving visual inputs uses computer vision to recognize and interpret these visuals, completing the
perception process. Conversely, NLP allows a machine to understand, interpret, and respond to language inputs,
ensuring accurate command execution.

Test Your Knowledge - 1

Multiple Choice Questions (MCQs)

1. What is Artificial Intelligence?
(a) Playing a game (b) Making a machine faster
(c) Making a machine intelligent (d) Programming with your intelligence
2. Who coined the term "Artificial Intelligence"?
(a) Charles Babbage (b) John McCarthy
(c) Dave Waters (d) Stephen Hawking
3. Which are the two basic components of NLP?
(a) Natural Language Understanding (b) Natural Language Generation
(c) Both (a) and (b) (d) None of these
Ans. 1. (c), 2. (b), 3. (c).
Short Answer Type Questions
1. Define the term AI.
Ans. Artificial Intelligence (AI) is the ability of a machine to think and learn. In other words, AI refers to the ability
of machine to machines to perform cognitive tasks like thinking, perceiving, learning, problem solving and
decision making; it is inspired by the ways people use their brains to perceive, learn, reason and decide the
action.

SESSION-1.2 AI PROJECT CYCLE

LEARNING OUTCOMES
• Identify the AI Project Cycle framework.

Introduction to AI Project Cycle

Developing an AI system requires a structured approach similar to any complex project. The AI project life cycle
outlines the key stages involved in building and deploying an AI solution. It typically includes the following stages:

Problem Data Data

Modelling Evaluation Deployment
Scoping Acquisition Exploration

1. Problem Scoping: Identifying and defining the problem that the AI system will address. This involves
understanding the requirements, objectives, and constraints of the project.
2. Data Acquisition: Gathering and collecting relevant data that will be used to train and validate the AI
model. This data could include images, text, videos, or any other form of structured or unstructured data.

6 AI REFLECTION, PROJECT CYCLE AND ETHICS

 3. Data Exploration: Exploring and analyzing the collected data to understand patterns, correlations, and
insights that can inform the development of the AI model.
4. Modeling: Developing and selecting appropriate AI models or algorithms that will be used to build the AI
system. This stage involves training the models on the collected data to make predictions or decisions.
5. Evaluation: Testing and evaluating the performance of the AI models to ensure they meet the desired
accuracy, reliability, and effectiveness criteria. This stage may involve refining the models and iterating
based on feedback.
6. Deployment: Implementing the AI model into the real-world environment where it will be used. This
involves integrating the model into existing systems or applications and making it available for end-users.
Let us understand this with the help of an example:
Obesity, a complex health condition characterized by excessive
body fat accumulation, has reached epidemic proportions globally. It is
a significant risk factor for a multitude of chronic diseases, including
heart disease, type 2 diabetes, stroke, and certain types of cancer. The
rising prevalence of obesity is attributed to various factors such as
sedentary lifestyles, unhealthy dietary habits, and genetic
predisposition. Early detection of obesity is crucial for preventing its
associated health complications. To address this challenge, data-driven
approaches and AI models are being developed to create more
accurate and personalized obesity prediction tools.
Now that you are aware of the AI concepts, write down what is the aim of the AI model for the given scenario.
___________________________________________________________________________________________
___________________________________________________________________________________________
Yes, the aim of this obesity prediction system is to accurately predict the likelihood of obesity in individuals. In
this stage, you identify and define the problem that the AI system will address. This involves understanding the
requirements, objectives, and constraints of the project. This is Problem Scoping.
As you proceed, you will start collecting the following data:
• Demographic information (age, gender, ethnicity)
• Health metrics (height, weight, BMI, blood pressure, cholesterol levels)
• Lifestyle factors (diet, exercise habits, smoking status)
• Medical history (family history of obesity, existing health conditions)
This data becomes the base of the obesity prediction system. Note that the data needs to be accurate and reliable
as it ensures the efficiency of your system. This is known as Data Acquisition.
After acquiring the required data, you realize that it is not uniform. There might be missing values,
inconsistencies, or outliers. To address these issues, you perform data cleaning and preprocessing. Once the data is
cleaned and prepared, you explore it to uncover patterns and relationships between variables. For example, we
might find that people who eat a lot of sugary drinks and don't exercise much are more likely to be overweight. This is
known as Data Exploration.
Following this, you write special computer programs called algorithms and then train models to learn patterns
between features and obesity. This is known as Modelling.
Then you evaluate it based on some performance criteria/metrics. For example, test it on a different group of
people to see how well it works. This is known as Evaluation. You notice that the model is 75% accurate. So, you
can improve this by changing the algorithm or by fine tuning some parameters used in the model.
Finally, when you are satisfied with the performance of the model, you put it for work in the form of an app for

R.K.- Artificial Intelligence - IX Part-B 7

everyone to use. This is known as Deployment. Multiple models are often developed and compared based on their
evaluation metrics. The model with the best overall performance is selected for deployment.

Did you know?

Voice assistants are being incorporated into a wide range of consumer products; nearly half of US adults
(46%) said they now use these applications to interact with smartphones and other devices. (Source: Pew
Research Centre)

AI Project Cycle Mapping

AI project cycle mapping involves aligning the specific actions and tasks undertaken in an AI project with the
broader stages of the AI project lifecycle. This process helps in:
• Understanding the project's progress: By mapping specific activities to lifecycle stages, teams can track
their progress and identify potential bottlenecks.
• Ensuring project completeness: Mapping helps verify that all essential phases of the AI project lifecycle
are covered.
• Facilitating communication: A clear mapping can improve communication among team members and
stakeholders.
Example: Obesity Prediction Model
Let us map the steps involved in developing an obesity prediction model to the AI project lifecycle stages:

Problem Scoping Data Acquiaition Data Exploration

Identifying obesity as Collecting data on Analyzing the
a growing demographics, collected data to
health concern health metrics, identify patterns,
and defining the goal lifestyle factors, and correlation, and
of predicting medical history. potential predictors
obesity risk. of obesity.

Modelling Evaluation Deployment

Selecting and Assessing the Using a mobile app
training machine model's to input user's data
learning algorithms performance using and receive
to build the metrics. predictions...
prediction model.

Match the following

Task AI Project Cycle Stage
Define the problem Problem scoping
Collect the data Data acquisition
Uncover patterns and relationships Data exploration
Write programs Modelling
Check the performance Evaluation
Put it to work Deployment

8 AI REFLECTION, PROJECT CYCLE AND ETHICS

Importance/Need for An AI Project Cycle
The AI project cycle is a structured approach to developing AI solutions. It breaks down the complex process of
AI development into smaller, more manageable steps, making it easier to track progress, identify potential issues,
and collaborate with other teams. This structured approach also improves the efficiency of the development process
and reduces the risk of errors or delays.
The AI project cycle typically consists of the following stages:
1. Problem definition: Clearly define the problem that the AI solution will address.
2. Data acquisition: Gather and prepare the data needed to train the AI model.
3. Data exploration: Analyse the data to identify patterns and trends.
4. Model development: Build and train the AI model.
5. Model evaluation: Test the model's performance and make improvements as needed.
6. Deployment: Deploy the model into a production environment.
By following these steps, organizations can increase their chances of successfully developing and deploying AI
solutions.
In addition to the benefits mentioned above, the AI project cycle also helps to

Enhance efficiency Enable Modularity Foster ethical considerations

By breaking down the process Each stage of the cycle can be Integrating ethical principles
into smaller, manageable stages, considered a module, allowing throughout the project ensures
the AI project cycle accelerates for independent development responsible and trustworthy AI
development and reduces time- and potential reuse of solutions.
to-market. components.

Test Your Knowledge - 2

Multiple Choice Questions (MCQs)

1. Problem scoping, data acquisition, data exploration are the stages of .................. .
(a) AI Project cycle (b) Problem scoping
(c) Domains of AI (d) None of the above
2. The third stage of project cycle is ..................... .
(a) Data Acquisition (b) Data Exploration
(c) Evaluation (d) Deployment
3. The last stage of the project cycle is .................... .
(a) Data acquisition (b) Data exploration
(c) Modelling (d) Deployment
Ans. 1. (a), 2. (b), 3. (d).
Short Answer Type Questions
1. Name the stages of AI project cycle.
Ans. The stages are problem scoping, data acquisition, data exploration, modelling, evaluation and deployment.
2. Why do we need AI Project cycle?
Ans. We need AI Project cycle to
• Enhance efficiency
• Enable modularity
• Foster ethical considerations

R.K.- Artificial Intelligence - IX Part-B 9

 SESSION-1.3 PROBLEM SCOPING

LEARNING OUTCOMES
• Learn problem scoping and ways to set goals for an AI project.
• Identify stakeholders involved in the problem scoped. Brainstorm on the ethical issues involved around
the problem selected.
• Understand the iterative nature of problem scoping for in the AI project cycle. Foresee the kind of data
required and the kind of analysis to be done.

What is Problem Scoping?

Problem scoping refers to the identification of a problem and the vision to solve it. In other words, the problem
scoping is defining of what is to be done to solve a problem and plan to be followed for solving the problem and plan
to be followed for solving the problem and ultimately what goals needs to be achieved on the completion of project.
Problem scoping is the very first stage of any project or task that we take on. Problem scoping involves
identifying the problem, setting of goals, identifying the stakeholders, identifying existing measures and identifying
the ethical measures.
Once problem scoping is done properly it gives a clear vision of the problem which is otherwise very abstract and
undefined.

Identifying Identifying
Identifying Setting of Identifying the
Existing the Ethical
the problem Goals Stakeholders
Measures Measures

Did you know?

The impact of AI technologies on business is projected to increase labour productivity by up to 40% and
enable people to make more efficient use of their time. (Source: Adobe)

TOPIC SELECTION AND REFINEMENT

1. Identifying a Theme
The first step in selecting a problem for AI solution is to identify a broad theme or domain of interest. This
could be based on your passion, expertise, or the current trends in the industry. Some potential themes
include:
• Healthcare: This encompasses a vast array of potential problems, from disease diagnosis to drug
discovery.
• Finance: Areas like fraud detection, algorithmic trading, and risk assessment offer opportunities for AI
applications.
• Environment: Climate change, natural disaster prediction, and resource management are pressing
environmental challenges.
• Education: Personalized learning, intelligent tutoring systems, and educational content creation are
potential areas of focus.

10 AI REFLECTION, PROJECT CYCLE AND ETHICS

 Theme:
Health
Care

Topic: Chronic Diseases Topic 2 Topic 3 Topic 4 Topic 5

Sub Topic 1: Sub Topic 2 Sub Topic 3 Sub Topic 4 Sub Topic 5
Obesity

Problem: Accurate and early prediction of obesity to enable timely interventions and improve public health
outcomes.

2. Choosing a Topic
Once you have selected a theme, narrow down your focus to a specific topic. This should be an area within
the theme where you see a significant problem or opportunity for AI to make a positive impact.
For example, within healthcare, chronic diseases represent a significant challenge and opportunity for AI
solutions.
3. Identifying Subtopics and Problems
Further refine your focus by selecting a subtopic within the chosen topic. This will help you define the specific
problem you want to address. For example, Obesity is a prevalent and complex chronic disease with far-
reaching health implications.

4Ws Problem Canvas

Problem scoping is a tedious task. It needs a lot of investigation by analysing the business process to address the
problem faced. For this, one needs to go through to the related documents, interview people, gather information
and discuss with the people involved. To effectively address a problem, it is essential to have a clear understanding of
its nature, impact, and context.
The 4W problem canvas is a framework that can help you in identifying the key elements. The 4W here means
Who? What? Where? and Why?

R.K.- Artificial Intelligence - IX Part-B 11

 Who What
• Identify stakeholders: • Define the problem:
Deter mine who is directly or Clearly articulate the issue at hand,
indirectly affected by the problem. including its scope and severity.
• Understand stakeholders: Research • Gather evidence: Collect data,
their needs, perspectives, and statistics, and case studies to support
priorities. the problem's existence.

Where Why
• Analyze context: • Assess impact: Evaluate the
Understand the environment in consequences of the problem on
which the problem occurs. individuals and society.
• Identify locations: • Define benefits:
Determine the geographical Articulate the potential value of a
distribution of the problem. solution.

Questions to ask are:

Who
• Who are the stakeholders?
• What do you know about them?

What
• What is the problem?
• How do you know that it is a problem (is there any evidence)?

Where
• What is the context/ situation the stakeholders experience regarding the problem?
• Where is the problem located?

Why
• Why will this solution be of value to the stakeholders?
• How will the solution improve their situation?

After filling the 4Ws Problem canvas, you now need to summarise all the answers into one template known as
the problem statement template. A problem statement template is a structured framework that
summarises the essence of a problem, including its impact, context, and potential solutions. It serves
as a foundation for further analysis and development. The format of the template is

Our [stakeholders] Who?

have/has a problem that [issue/problem/need] What?

When/while [context/situation]. Where?

An ideal solution would be [benefit of solution]. Why?

12 AI REFLECTION, PROJECT CYCLE AND ETHICS

Problem Statement Template for Obesity Prediction System

Our people Who?

have/has a problem that there is a risk of obesity What?

When/while opting for unhealthy lifestyle choices or genetic Where?

predisposition, or environmental factors.

An ideal solution would be to develop an AI-powered obesity prediction Why?

model to identify individuals at risk.

Test Your Knowledge - 3

Multiple Choice Questions (MCQs)

1. Arrange the following in correct order:
(i) Identify a theme for the problem (ii) Identify the subtopic
(iii) Choosing a topic (iv) Identify the problem
(a) ii, i, iii, iv
(b) i, ii, iv, iii
(c) iv, i, iii, ii
(d) i, iii, ii, iv
2. Which of the following block speaks about the solution to the problem?
(a) Who (b) Where
(c) Why (d) What
Ans. 1. (d), 2. (c).
Short Answer Type Questions
1. Who are the stakeholders in the problem scoping stage?
Ans. Stakeholders are the people who are getting affected by the problem directly or indirectly.
2. Why is there a need to use a Problem Statement Template during problem scoping?
Ans. A problem statement template is a structured framework that summarises the essence of a problem,
including its impact, context, and potential solutions. It serves as a foundation for further analysis and
development.

SESSION-1.4 DATA ACQUISITION AND DATA EXPLORATION

LEARNING OUTCOMES
• Identify data requirements and find reliable sources to obtain relevant data
• Understand the purpose of Data Visualisation
• Use various types of graphs to visualise acquired data

Data acquisition is the second stage of the AI project cycle. It involves collecting, cleaning, and preparing data
essential for model development. Data comprises information, facts, or statistics that are crucial for training and
validating AI models. The quality and reliability of this data are critical to ensure the effectiveness and accuracy of the
AI system.

R.K.- Artificial Intelligence - IX Part-B 13

14 AI REFLECTION, PROJECT CYCLE AND ETHICS
R.K.- Artificial Intelligence - IX Part-B 15
Data Visualisation
Data exploration is the process of examining data to discover patterns, trends, and anomalies. It involves
using statistical methods and visualization techniques to understand the data's characteristics and identify potential
relationships. This initial investigation helps guide subsequent analysis and modeling efforts.
Importance of Data Exploration:
1. Understanding the Data: Gain insights into data characteristics and distributions.
2. Identifying Patterns: Discover relationships between variables and potential correlations.
3. Detecting Outliers: Identify unusual data points that may require further investigation.
4. Feature Engineering: Create new variables or transform existing ones to improve model performance.
5. Data Cleaning: Identify and address data quality issues such as missing values and inconsistencies.
Data Visualization Techniques:
To effectively explore data, we often rely on visual representations. Visit this link: https://datavizcatalogue.com/
to explore various visual representations. Some of the common charts are:

Bar Chart Box and Whisker plot Pie Chart

Histogram Scatter Chart

16 AI REFLECTION, PROJECT CYCLE AND ETHICS

Identify the following charts:

Data visualisation tools help the designers to create visual representation of the data sets in an easy manner
which then can be easily understood. Some of these tools are as follows—
1. Spreadsheet Package. Spreadsheet Packages such as MS-Excel, LibreofficeCalc, Google Sheets provide
a variety of graphical tools to represent data. These tools re suitable for working with simple data only.
2. Tableau. Tableau is the widely used data visualisation tool. It is simple to use but powerful tool for creating
interactive graphs and charts in the form of dashboards and worksheets to gain business insights. It is
capable of handling enormous, frequently updated data sets.
3. Candela. Candela is an open source tool to create rich data visualisations. It provides a library of charts
graphs and plots to integrate with your data sets.
4. Google Data Studio. It is a dashboard and reporting tool that allows us to create appealing and
informative reports using the data sets available. It is easy to use, customise and even allows to share the
work done.
5. Quick View. Quick View is used to analyse data for decision making. This tool is known for its capability of
customisation and extensive features.

Did you know?

Consumers use more AI than they realise. While only 33% think they use AI-enabled technology, 77%
actually use an AI-powered service or device. (Source: Pega)

R.K.- Artificial Intelligence - IX Part-B 17

 When you visualize the obesity dataset, you can note that obesity runs in the family. All those categorised as
overweight or obese had family members suffering from weight problems.
350 family_history_with_overweight
yes
300 no

250

200
Count

150

100

50

0
Insufficient_Weight Normal_Weight Overweight_Level_I Overweight_Level_II Obesity_Type_I Obesity_Type_II Obesity_Type_III
NObeyesdad

Test Your Knowledge - 4

Multiple Choice Questions (MCQs)

1. Which of the following are valid sources for collecting data? [CBSE]
(i) Announcements (ii) Surveys
(iii) Web scraping (iv) Application Programming Interface
(a) (i) , (ii) and (iii) (b) (i) , (ii) and (iv)
(c) (ii) , (iii) and (iv) (d). (i) , (iii) and (iv)
2. Identify the correct representations in a system map [CBSE]
(a) Circles -> elements; arrowed lines -> relationship
(b) Squares -> elements ; dotted lines -> relationship
(c) Circles -> elements ; dotted lines -> relationship
(d) Squares -> elements ; arrowed lines -> relationship
3. Which of these is NOT used for visualization in AI? [CBSE]
(a) Sketchy Graph (b) Scatter plot
(c) Ruler (d) Bar chart
Ans. 1. (c), 2. (a), 3. (c)
Short Answer Type Questions
1. Differentiate between training data and testing data.
Ans. Training data Testing data
Is the one on which we train and fit our model Is used only to assess performance of model
basically to fit the parameters
Output is available to model The unseen data for which predictions have to be
made

2. Why we need to explore data?

Ans. We need to explore data to
• Get a sense of trends, relationships and patterns present in data
• Decide which model to use in our AI Project cycle
• Easier to comprehend information
• Easier to communicate the story with others

18 AI REFLECTION, PROJECT CYCLE AND ETHICS

 SESSION-1.5 MODELLING

LEARNING OUTCOMES
• Understand modelling (Rule based and Learning-based)

Artificial Intelligence (AI) is the broader field that encompasses the

Artificial
development of intelligent agents, which are systems that can reason, learn, and
Intelligence
act autonomously.
Machine Learning (ML) is a subset of AI that focuses on enabling Machine
systems to learn from data without being explicitly programmed. ML Learning
algorithms identify patterns in data and use them to make predictions or
decisions.
Deep Learning (DL) is a specialized subset of machine learning that Deep
utilizes artificial neural networks with multiple layers to analyze complex Learning
patterns in large datasets. Deep learning has been instrumental in advancements
such as image recognition, natural language processing, and speech recognition.
While human perception relies heavily on visual representations to understand data
trends and patterns, machines excel at processing numerical data. This fundamental difference necessitates a
translation between the human-readable visualizations and the machine-interpretable numerical format. The core
of every AI model lies in its ability to mathematically describe the relationships between parameters. Therefore,
when we discuss developing AI models, we are referring to the mathematical approach used for data analysis. AI
models can be classified as follows:

Classification
algorithm
Supervised
learning
Regression
algorithm
Machine
Learning Unsupervised Clustering
Learning based learning algorithm

AI Models Deep Learning

Reinforcement
Rule based
learning

AI systems are primarily categorized into two types: rule-based and learning-based systems, each with distinct
characteristics and applications.
A. Traditional Rule-Based Systems operate on a set of predefined rules and logic, similar to following a
detailed instruction manual. When these systems encounter a problem, they consult their established rules
to determine the appropriate decision.
• One of the main strengths of rule-based systems is their transparency and explainability; the decision-
making process is clear and straightforward, making it easy to understand how conclusions are reached.
Additionally, these systems are highly efficient in domains where rules are well-defined and consistent.
• However, rule-based systems also have notable limitations. They are rigid and inflexible, struggling to
adapt to new or unexpected situations. This rigidity makes them less effective when dealing with
complex problems involving multiple variables.

R.K.- Artificial Intelligence - IX Part-B 19

 Did you know?
Business leaders said they believe AI is going to be fundamental in the future. In fact, 72% termed it a
“business advantage.” (Source: Adobe)

B. Learning based systems learn from data and improve their performance over time. They identify
patterns and relationships within data to make predictions or decisions. They can be further divided into two
main types: machine learning and deep learning. Machine learning systems learn from data without
needing explicit programming for each task, while deep learning, a subset of machine learning, utilizes
artificial neural networks with multiple layers to analyze complex patterns in data.
The strengths of learning-based systems include –
• Adaptability: Can handle complex and evolving problems.
• High accuracy in pattern recognition tasks.
• Continuous improvement through learning.
However, they also come with limitations. Learning-based systems often require large amounts of data to
function effectively and can be computationally expensive due to the resources needed for processing and learning.
Furthermore, these systems can lack transparency in their decision-making processes, making it harder to
understand how decisions are reached.

Supervised Learning
Supervised learning is a core concept in machine
learning where a model is trained on labeled data,
which means the data comes with predefined correct
answers. This labeled data is used to teach the model
to make predictions on new, unseen data. Labels
assist in linking input data to the correct output.
Supervised learning models learn to associate these
input data points with their respective output labels.
There are two main types of problems that utilize
supervised learning algorithms: classification and
regression. Regression algorithms predict a
continuous numerical value, while classification
algorithms predict a categorical value. Common supervised learning algorithms include linear regression, K-Nearest
Neighbors (KNN), decision trees, support vector machines (SVM), random forests, and neural networks.
The following are some of the real-world scenarios where classification is used:
• Email Spam Filtering: Determining whether an email is spam or not.
• Medical Diagnosis: Classifying diseases based on patient symptoms and test results.
• Customer Churn Prediction: Predicting whether a customer will discontinue a service.
The following are some of the real-world scenarios where regression is used:
• Housing Price Prediction: Estimating the price of a house based on features like size, location, and
number of bedrooms.
• Sales Forecasting: Predicting future sales based on historical data and market trends.
Unsupervised Learning
In Unsupervised Learning the data used to train the machine are neither labelled nor classified. The machine
learning model (algorithm) discover the patterns on its own. The goal is to organize unstructured information based

20 AI REFLECTION, PROJECT CYCLE AND ETHICS

on similarities, patterns, and differences without any
prior training or guidance. In this type of learning, the
machine is tasked with finding hidden structures in
the unlabelled data without supervision.
Clustering is a common unsupervised learning
technique used to group a set of objects so that
objects in the same group (or cluster) are more
similar to each other than to those in other groups.
Some well-known unsupervised learning algorithms
include k-means clustering, hierarchical clustering,
and principal component analysis (PCA).
The following are some of the real-world
scenarios where clustering is used:
• Customer Segmentation: Grouping customers based on similar purchasing behaviour or demographics.
• Document Clustering: Organizing documents into groups based on content similarity.
• Image Segmentation: Grouping pixels in an image into different regions based on color, texture, or other
features.

Reinforcement Learning
Reinforcement learning (RL) is a machine learning
approach where an agent learns to make decisions by
taking actions in an environment to maximize
cumulative reward. The agent learns from the
outcomes of its actions through trial and error without
being given examples of correct input-output pairs. For
instance, in a game where a robot collects a diamond
while avoiding obstacles, the robot tries different paths
and learns to choose the path that leads to the diamond
with the fewest obstacles, accumulating points for
correct steps and losing points for incorrect ones.
Common applications of RL include gaming,
autonomous vehicles, and personalized treatment
plans, with popular algorithms such as Deep Q-
networks (DQN).

Identify which of the following are examples of classification/regression/clustering. [CBSE]

a. Making a diagnosis for a patient on the basis of their symptoms
b. Price prediction for a house coming up on sale
c. HR shortlisting applications for a specific position in an interview
based on information provided in candidates' resume
d. Credit Card Fraud prevention
e. SPAM filters

For our Obesity prediction model, we have labeled data (patients with their obesity status), and so supervised
learning is the most suitable approach. Algorithms such as Logistic Regression/Decision trees could be used.

R.K.- Artificial Intelligence - IX Part-B 21

 Test Your Knowledge - 5

Multiple Choice Questions (MCQs)

1. ___________refers to developing algorithms, also called models which can be trained to get intelligent
outputs.
(a) Problem Scoping (b) Evaluation
(c) Modelling (d) Data Acquisition
2. Which of the following approaches have the rules predefined by the developer?
(a) Rule based approach (b) Learning based approach
(c) Machine learning approach (d) Robot learning approach
Ans. 1. (c), 2. (a).
Short Answer Type Questions
1. How is Machine Learning related to Artificial Intelligence?
Ans. Machine Learning (ML) is a subset of AI that specifically focuses on enabling machines to learn from data
which may or may not involve explicit programming.

SESSION-1.6 EVALUATION AND DEPLOYMENT

LEARNING OUTCOMES
• Understand various evaluation techniques.
• Understand the importance of deploying AI models into real-world applications.

Data Evaluation is a critical step in the AI project lifecycle that determines the effectiveness of a model. It
involves assessing the model's performance on a separate dataset called the testing dataset. This dataset was not
used during the training phase, ensuring an unbiased evaluation.
Key Evaluation Metrics
To measure model performance, several metrics are used:
• Accuracy: The proportion of correct predictions out of the total predictions.
• Precision: The proportion of positive predictions that were actually correct.
• Recall: The proportion of actual positives that were correctly identified.
• F1-score: A harmonic mean of precision and recall, providing a balanced measure.
Model Evaluation Terminologies
Model evaluation is an important step in the machine learning process. It involves checking how well a model
works by comparing its predictions to real-world data. The goal is to see how accurate, reliable, and useful the model
is for a specific task. Two key ideas are involved in this process: prediction and reality.
Prediction: This is what the model guesses or estimates when it is given new data it hasn't seen before. The
model makes these predictions based on the patterns it learned during its training.
Reality: This is the actual or correct outcome for the data, also known as the ground truth. It serves as the
standard to which the model's predictions are compared. By comparing the model's predictions to the actual results,
we can measure how well the model is performing and identify areas where it might need improvement. This
comparison is the foundation for various methods used to evaluate the quality of the model.
There are various new terminologies which come into the picture when we work on evaluating our model.
Let us explore them with an example.

22 AI REFLECTION, PROJECT CYCLE AND ETHICS

 In a village where inconsistent weather conditions frequently disrupt school schedules, causing inconvenience
and potential safety risks for students, an AI-powered model has been implemented to predict school closures due to
rain. This model utilizes inputs such as real-time and historical weather data, including temperature, humidity,
rainfall, and wind speed, along with geographic data like location-specific information on altitude, terrain, and
drainage systems. Additionally, the model considers school infrastructure details, including building conditions,
drainage systems, and emergency preparedness. By analyzing these inputs, the model outputs a binary classification
indicating whether schools should be closed or remain open, helping to ensure student safety and minimize
disruption.
Solution: Develop an AI-powered model to predict if schools should be closed due to rain.

PREDICTION: YES REALITY : YES

TRUE POSITIVE

True Positive (TP): The model correctly predicts school closure on a rainy day.

PREDICTION : NO REALITY : NO

TRUE NEGATIVE

True Negative (TN): The model correctly predicts and the school remains open on a clear day.

R.K.- Artificial Intelligence - IX Part-B 23

 PREDICTION: YES REALITY : NO

FALSE POSITIVE

False Positive (FP): The model incorrectly predicts school closure on a clear day.

PREDICTION: NO REALITY : YES

FALSE NEGATIVE

False Negative (FN): The model incorrectly predicts and the school remains open on a rainy day

24 AI REFLECTION, PROJECT CYCLE AND ETHICS

 • True Positive
TP • Prediction and Reality matches? Yes (True)
• Prediction is Yes (Positive) Did you know?
Forty-seven per cent of digitally mature
organisations or those that have
advanced digital practices, said they have
• True Negative a defined AI strategy. (Source: Adobe)
TN • Prediction and Reality matches? Yes (True)
• Prediction is No (Negative)

• False Positive
FP • Prediction and Reality matches? No (False)
• Prediction is Yes (Positive)

• False Negative
FN • Prediction and Reality matches? No (False)
• Prediction is No (Negative)

Once the models are ready, it's crucial to assess their performance. To
determine which algorithm yields the most accurate predictions, we will
evaluate each model's performance. One key metric for evaluating
classification models is the Receiver Operating Characteristic (ROC) curve.
The ROC curve graphically represents a model's ability to distinguish
between positive and negative classes. By comparing the ROC curves of our
different algorithms, we can identify the model that demonstrates the best
overall performance. The accompanying figure provides a visual
comparison of the accuracy of the three algorithm samples under
consideration.
Our Obesity prediction models’ performance can be measured using
metrics like accuracy, precision, recall, and F1-score. It can then be fine-
tuned based on evaluation results.
Deployment
Deployment is the final stage in an AI project, where the developed model is integrated into a real-world
environment. Deployed AI project can be used on Mobile Applications, Website Applications, etc., Some of the key
steps in deployment process include:

Integration Monitoring
Testing and
with existing and
Validation
systems Maintenance

R.K.- Artificial Intelligence - IX Part-B 25

 1. Testing and Validation
Before deployment, the AI model undergoes rigorous testing and validation to ensure its accuracy, reliability,
and performance. This involves:
• Evaluating the model's performance on a diverse dataset to identify potential biases or errors.
• Conducting sensitivity analysis to assess the model's robustness to changes in input data.
• Verifying that the model meets predefined performance metrics and business objectives.
2. Integration with Existing Systems
Integrating the AI model into existing systems and workflows is crucial for seamless operation. Key
considerations include:
• Designing APIs or interfaces for model interaction.
• Ensuring compatibility with data formats and structures.
• Developing data pipelines for efficient data flow between the model and other systems.
• Considering security and privacy implications during integration.
3. Monitoring and Maintenance
Continuous monitoring and maintenance are essential to ensure the model's ongoing performance and
effectiveness. This involves:
• Tracking key performance indicators (KPIs) to measure model accuracy and efficiency.
• Implementing alert systems to detect anomalies or performance degradation.
• Regularly retraining the model with new data to maintain accuracy.
• Addressing model drift and concept drift to prevent performance decline.
• Conducting regular audits and evaluations to assess the model's impact.
AI Project Cycle Mapping Template
Once the Obesity prediction model is evaluated, it can be integrated into a healthcare system or you can develop
a user-friendly application. It can also provide recommendations for users based on the prediction. Let us map the
Obesity prediction model to AI Project Cycle Mapping Template:

Data Data
Modelling Evaluation Deployment
Acquisition Exploration

Collect a Analyze the Develop and train Evaluate model Integrate the
dataset dataset to machine learning performance using model into a
containing understand models (e.g., metrics like healthcare
patient records correlations logistic regression, accuracy, system or
with attributes between obesity decision trees, precision, recall, develop a user-
like, age, and different random forest) and F1-score. friendly
gender, height, factors. Identify to predict obesity Fine-tune the application for
weight, potential features risk based on model based on obesity risk
blood pressure, for model the collected data. evaluation assessment.
cholesterol training. Handle results. Provide
levels, and missing values recommendations
dietary habits, and outliers. for users based
on the
prediction.

26 AI REFLECTION, PROJECT CYCLE AND ETHICS

 Test Your Knowledge - 6

Multiple Choice Questions (MCQs)

1. In the context of the school closure prediction model, a False Negative (FN) occurs when:
(a) The model correctly predicts school closure on a rainy day.
(b) The model incorrectly predicts school closure on a clear day.
(c) The model correctly predicts and the school remains open on a clear day.
(d) The model incorrectly predicts and the school remains open on a rainy day.
2. In which step of the deployment process do you ensure the model's ongoing performance and effectiveness?
(a) Testing and validation of the AI model
(b) Integration of the model with existing systems
(c) Monitoring and maintenance of the deployed model
(d) Data acquisition
Ans. 1. (d), 2. (c).
Fill in the Blanks
1. _____________ is the final stage in the AI project cycle where the AI model or solution is implemented in a
real-world scenario.
Ans. 1. Deployment.

SESSION-1.7 ETHICS AND MORALITY

LEARNING OUTCOMES
• Differentiate between morality and ethics.
• Identify ethical concerns related to personal data.
• Understand ethical challenges and principles in AI.
• Analyse ethical implications of AI in real-world scenarios.

Scenario 1: Finding a Lost Wallet

You find a wallet on the street containing a significant amount of
cash and identification. What do you do?
______________________________________________________
______________________________________________________
______________________________________________________
______________________________________________________
This situation primarily deals with honesty. Honesty is a moral
value, a personal belief about what is right. Returning the wallet is
the honest thing to do.
Morals are personal beliefs about right and wrong, often influenced by upbringing,
culture, and religion. They guide individual behavior.
Scenario 2: The Damaged Product
You purchase a new electronic device that malfunctions shortly after purchase. The store offers you
a replacement, but you suspect they might repair the faulty device and resell it. What do you do?

R.K.- Artificial Intelligence - IX Part-B 27

______________________________________________________________________________________________
This situation primarily deals with ethics. The company's policies and consumer rights (ethics) come into play.
Ethics are a system of moral principles that govern a group or society. They are often codified into
rules, laws, or professional codes of conduct.

Ethics vs Morals

Morals Ethics

Morals primarily focus on the individual. Ethics are concerned with societal norms.

They are derived from personal beliefs, upbringing, They stem from shared values and community
and culture. agreements.

They tend to be subjective, reflecting personal They are considered more objective, based on broader
perspectives. societal standards.

They function as an internal compass guiding They provide an external framework for conduct.
individual behaviour.

Examples of morals include honesty, fairness, and Examples of ethics encompass professional codes,
compassion. laws, and human rights.

Let us Identify: Morals or Ethics?

You have always wanted a pet. One sunny afternoon, while walking down the street, you come across a lost dog. It is
a charming, friendly and it seems to have taken a liking to you. After much deliberation, you decide to take it home
and care for it.
______________________________________________________________________________________________
A hospital faces a critical shortage of ventilators during a pandemic. With an overwhelming number of patients in
need, the medical team is forced to make decisions about who receives life-saving treatment. The chief medical
officer develops a system based on factors such as age, underlying health conditions, and the likelihood of patient
recovery.
______________________________________________________________________________________________

Moral Machine
Ethical dilemmas arise when making choices becomes difficult due to conflicting moral principles. In the
world of Artificial Intelligence (AI), these dilemmas become especially complex due to the powerful capabilities and
potential impacts of AI systems on individuals, society, and the environment. Addressing these ethical issues is
critical for responsible development and deployment of AI. The Moral Machine (https://www.moralmachine.net/),
created by researchers at MIT, offers an interactive platform to explore ethical dilemmas in AI.
The Moral Machine presents users with hypothetical scenarios where autonomous vehicles must make split-
second decisions that could impact human lives. These scenarios often involve moral conflicts, such as deciding
whether to prioritize the safety of passengers or pedestrians, elderly individuals versus children, or obeying traffic
laws versus avoiding greater harm.

28 AI REFLECTION, PROJECT CYCLE AND ETHICS

 The Moral Machine underscores the importance of integrating ethics into AI development. As AI systems
become more autonomous and pervasive in society, they will inevitably face situations where ethical principles must
guide their actions. Developers must consider how to program AI to make decisions aligned with societal values,
ensuring transparency and accountability in their algorithms. It also stresses on the fact that Public discourse plays a
pivotal role in establishing norms and standards that uphold ethical practices in AI development and deployment.
Ethics and Personal Data
Ethics and Personal Data closely connected. As technology continues to grow, the collection, storage and use of
personal data are becoming more common. With this increase in data handling, it's essential to have a strong ethical

R.K.- Artificial Intelligence - IX Part-B 29

framework in place to protect people's privacy, prevent misuse of their information, and ensure that data is managed
responsibly.
Yash, a data analyst at a prominent corporation, is deeply immersed in the
digital ecosystem. His daily work revolves around managing sensitive corporate
and client data, heavily reliant on his laptop and smartphone. He leverages AI-
powered tools to streamline tasks such as data analysis, visualization, and
report generation. These tools enable him to efficiently create complex
PowerPoint presentations and interactive dashboards. However, this extensive
use of technology also exposes him to potential risks, such as data breaches and
privacy infringements.

Vaishnavi, a school student, frequently uses smartphones and AI

applications to help with her homework and projects. Although AI provides
convenience and efficiency, it also brings certain challenges. For example,
generative AI tools like ChatGPT can sometimes produce incorrect or misleading
information, highlighting the importance of critical thinking and fact-checking.

Principles in AI Ethics
Principles of AI are a set of guidelines that govern the design, development, and Human
Inclusion
Rights
deployment of artificial intelligence systems. These principles aim to ensure that AI is
developed and used ethically, responsibly, and beneficially for society. To create AI
Privacy Bias
systems that benefit society, we must prioritize ethical considerations. Four key
principles guide the development and deployment of responsible AI:
• Human Rights • Bias • Privacy • Inclusion

Principle Meaning Explanation Example

Respecting AI systems should uphold human Surveillance systems should

Human
fundamental rights, like privacy, freedom of not excessively infringe on
Rights
human entitlements expression, and job displacement. personal liberties.

Unfair preference Facial recognition systems

Avoiding prejudice in AI algorithms
Bias for one group over should be accurate across
and data.
another different demographics.

Protecting Implementing robust

Safeguarding user data from
Privacy personal security measures to
unauthorized access and misuse.
information prevent data breaches.

Designing AI systems that are

Ensuring AI Making AI accessible and beneficial
Inclusion usable by people with
benefits everyone to all, regardless of background.
disabilities.

30 AI REFLECTION, PROJECT CYCLE AND ETHICS

 Did you know?
A recent World Economic Forum (WEF) survey revealed a startling truth: misinformation and
disinformation risks associated with Artificial Intelligence (AI) ranked as the second most concerning short-
term threat.

Match the Following

A social media platform collects user data without explicit consent. Human Rights
A visually impaired individual is unable to use a popular smartphone app. Bias
An AI- powered hiring tool consistently favors male candidates over female
Privacy
candidates.
Rapid advancements in AI and automation are displacing workers across industries,
Inclusion
leading to rising unemployment rates and exacerbating economic inequality.

Test Your Knowledge - 7

Multiple Choice Questions (MCQs)

1. The guiding principles to decide what is good or bad is known as:
(a) Ethics (b) Morals (c) Values (d) Norms
2. When a social media platform collects user data without explicit consent, it violates the principle of:
(a) Human Rights (b) Bias (c) Privacy (d) Inclusion
3. Artificial Intelligence and machine learning systems can display unfair behavior if not trained properly. This
is an example of:
(a) Privacy breach (b) Bias
(c) Human Rights violation (d) Lack of inclusion
Ans. 1. (a), 2. (c), 3. (b).
Short Answer Type Questions
1. Define principles of AI.
Ans. Principles of AI are a set of guidelines that govern the design, development, and deployment of artificial
intelligence systems. These principles aim to ensure that AI is developed and used ethically, responsibly, and
beneficially for society.

RECAP
• Artificial Intelligence (AI) is the branch of computer science that involves creating machines capable of
performing tasks that typically require human intelligence, such as learning, reasoning, problem-solving, and
decision-making.
• AI has become a part of our daily lives. We see a lot of applications of AI in everyday life.
• Based on the input data fed into AI, it can be classified into three domains—
1. Data 2. Computer Vision 3. Natural Language Processing (NLP)
• An AI project undergoes a set of 6 stages namely Problem scoping, data acquisition, data exploration,
modelling, evaluation and deployment.
• Problem scoping refers to the identification of a problem and the vision to solve it.
• The 4W problem canvas is a framework that can help you in identifying the key elements related to a problem.
The 4W here means Who? What? Where? and Why?

R.K.- Artificial Intelligence - IX Part-B 31

 • A problem statement template is a structured framework that summarises the essence of a problem, including
its impact, context, and potential solutions. It serves as a foundation for further analysis and development.
• Data comprises information, facts, or statistics that are crucial for training and validating AI models.
• The data is divided into two main subsets:
• Training Data: This subset is used to train the AI model, allowing it to learn and identify patterns.
• Testing Data: This subset is used to evaluate the model’s performance and accuracy, ensuring it
generalizes well to new, unseen data.
• A system map is a visual representation that shows the components and boundaries of a system at a specific
point in time. It helps us understand the relationships between these components and how they interact with
each other to produce a specific outcome.
• Data exploration is the process of examining data to discover patterns, trends, and anomalies. It involves using
statistical methods and visualization techniques to understand the data's characteristics and identify potential
relationships.
• The three key terminologies
1. Artificial Intelligence (AI) is the broader field that encompasses the development of intelligent agents,
which are systems that can reason, learn, and act autonomously.
2. Machine Learning (ML) is a subset of AI that focuses on enabling systems to learn from data without
being explicitly programmed. ML algorithms identify patterns in data and use them to make predictions or
decisions.
3. Deep Learning (DL) is a specialized subset of machine learning that utilizes artificial neural networks
with multiple layers to analyze complex patterns in large datasets.
• AI systems are primarily categorized into two types: rule-based and learning-based systems, each with distinct
characteristics and applications.
• Traditional Rule-Based Systems operate on a set of predefined rules and logic, similar to following a detailed
instruction manual.
• Learning based systems learn from data and improve their performance over time. They identify patterns and
relationships within data to make predictions or decisions.
• Supervised learning is a core concept in machine learning where a model is trained on labeled data, which
means the data comes with predefined correct answers.
• There are two main types of problems that utilize supervised learning algorithms: classification and regression.
Regression algorithms predict a continuous numerical value, while classification algorithms predict a
categorical value.
• In Unsupervised Learning the data used to train the machine are neither labelled nor classified. The machine
learning model (algorithm) discover the patterns on its own.
• Clustering is a common unsupervised learning technique used to group a set of objects so that objects in the
same group (or cluster) are more similar to each other than to those in other groups.
• Reinforcement learning (RL) is a machine learning approach where an agent learns to make decisions by
taking actions in an environment to maximize cumulative reward.
• Evaluation determines the effectiveness of a model. It involves assessing the model's performance on a
separate dataset called the testing dataset.
• Four key terminologies are used here: True Positive, False Positive, True Negative and False Negative.
• The ROC curve graphically represents a model's ability to distinguish between positive and negative classes.
• In deployment, the developed model is integrated into a real-world environment as a mobile application or a
web application.
• Morals are personal beliefs about right and wrong, often influenced by upbringing, culture, and religion. They
guide individual behavior.
• Ethics are a system of moral principles that govern a group or society. They are often codified into rules, laws,
or professional codes of conduct.

32 AI REFLECTION, PROJECT CYCLE AND ETHICS

 • Ethical dilemmas arise when making choices becomes difficult due to conflicting moral principles.
• Principles of AI are a set of guidelines that govern the design, development, and deployment of artificial
intelligence systems.
• Four key principles guide the development and deployment of responsible AI: Human Rights, Bias , Privacy ,
Inclusion

Objective Types Questions

?
?
Multiple Choice Questions (MCQs)
1. Which one of the following is an application of AI? [CBSE]
(a) Remote controlled Drone (b) Self-Driving Car
(c) Self-Service Kiosk (d) Self-Watering Plant System
2. What does NLP stand for in AI? [CBSE]
(a) Neutral Learning Projection (b) Neuro-Linguistic Programming
(c) Natural Language Processing (d) Neural Logic Presentation
3. This language is easy to learn and is one of the most popular languages for AI today: [CBSE]
(a) C++ (b) Java (c) Ruby (d) Python
4. How many stages are there in the AI Project cycle?
(a) 4 (b) 5 (c) 3 (d) 6
5. Which of the following comes under Problem Scoping?
(a) System Mapping (b) 4Ws Problem Canvas
(c) Data Features (d) Web scraping
6. Which canvas block think about the benefits which the stakeholders would get from the solution?
(a) Why (b) Where (c) When (d) What
7. Which of the following is not a visualization tool?
(a) Pie chart (b) Histogram (c) Paragraph (d) Scatter plot
8. Regression works with
(a) Intermittent data (b) Step function data
(c) Continuous data (d) Partially linear data
9. In which stage do we use the parameters accuracy, precision, recall and F1 score?
(a) Data exploration (b) Evaluation (c) Modelling (d) Data acquisition
10. Designing AI systems that are usable by people with disabilities – is an example of the AI Principle
(a) Human Rights (b) Bias (c) Privacy (d) Inclusion
Ans. 1. (b), 2. (c), 3. (d), 4. (d), 5. (b), 6. (a), 7. (c), 8. (c), 9. (b), 10. (d).
Fill in the Blanks
1. The full form of AI is ______________________.
2. ____________ is known as the father of AI.
3. _____________ domain enables computers to identify and process images as humans do. [CBSE]

AI REFLECTION, PROJECT CYCLE AND ETHICS 33

 4. Under the __________________stage of the AI Project Cycle, we look at various parameters which affect the
problem we wish to solve.
5. __________ helps us to summarise all the key points into one single Template so that in future, whenever
there is need to look back at the basis of the problem, we can take a look at the this and understand the key
elements of it.
6. _____________websites are used for data collection.
7. __________ refer to the type of data you want to collect.
8. ____________uses neural networks.
9. A ___________ approach is one in which data and rules are fed to the machine, and the machine reacts
accordingly to deliver the desired output.
10. ___________is the process of understanding the reliability of any AI model, based on outputs by feeding test
dataset into the model and comparing with actual answers.
11. Deployment as the ___________stage in the AI project cycle where the AI model or solution is implemented
in a real-world scenario.
12. The guiding principles to decide what is good or bad is known as ___________.
13. “Preference for one over the other” is known as ____________.
Ans. 1. Artificial Intelligence, 2. Dr. John Mc Carthy, 3. Computer Vision, 4. Problem Scoping, 5. Problem
Statement Template, 6. Open-sourced, 7. Data features, 8. Deep learning, 9. Rule based, 10. Evaluation,
11. final / last, 12. Ethics, 13. Bias.
State whether the given statements are True Or False:
1. Statistical Data refers to statistical techniques to analyse, interpret and draw insights from numerical/tabular
data.
2. The “Who” block helps you in analysing the people getting affected directly or indirectly due to it.
3. Data Acquisition is the third stage of AI Project cycle.
4. In a system map, if an arrow goes from X to Y with a – (minus) sign, it means that if X increases, Y decreases.
5. Machine learning is a subset of Deep learning.
6. A learning approach is one in which the machine is fed with data and the desired output, to which the
machine designs its own algorithm.
7. When prediction = Yes, and Reality = Yes, the model terminology used is False Positive.
8. Evaluation is the final stage in the AI project cycle where the AI model or solution is implemented in a real-
world scenario.
9. Morals are the beliefs dictated by our society.
10. An Ethical AI framework makes sure that transparency, fairness and accountability is developed into the
systems to provide unbiased results.
11. Artificial Intelligence and machine learning systems can display unfair behaviour if not trained properly.
12. AI can help in reducing human errors.
13. AI applications in healthcare can only assist with research and analysis.
14. AI can function effectively without any data.
Ans. 1. True , 2. True, 3. False, 4. True, 5. False, 6. True, 7. False, 8. False, 9. True, 10. True, 11. True, 12. True,
13. False, 14. False.

34 AI REFLECTION, PROJECT CYCLE AND ETHICS

 Match the Following
Column I Column II

1. Domain of AI (a) Continuous data

2. Data + ___ = AI (b) Statistical Data

3. AI Project cycle (c) static

4. Where block (d) algorithm

5. Data source (e) Deployment

6. Histogram (f) 6 stages

7. Rule based approach (g) guiding principles to decide what is good or bad

8. Last stage of AI Project cycle (h) context/situation/location of the problem

9. Ethics (i) preference for one over the other

10. Bias (j) Web scraping

Ans. 1. (b), 2. (d), 3. (f), 4. (h), 5. ( j), 6. (a), 7. (c), 8. (e), 9. (g), 10. (i).
Case Based Questions
1. Imagine two friends, Ashwat and Anurag, are deciding who will go first in a game they’re about to play. They
want a quick and fair way to make the decision. Ashwat suggests a simple game where they each choose one
of three options: A fist, An open hand, A fist with the index and middle fingers extended. They reveal their
choices simultaneously. They follow these outcome rules to determine the winner:
• A fist crushes an open hand
• An open hand covers a fist with the index and middle fingers extended
• A fist with the index and middle fingers extended cuts a fist
Ans. Rock, Paper, Scissors.
2. After the pandemic, it has been essential for everyone to wear a mask. However, you see many people not
wearing masks when in public places. Which domain of AI can be used to build a system to detect people not
wearing masks? [CBSE]
Ans. Computer Vision.
3. At this stage, you try to interpret some useful information out of the data you have acquired. For this, you
explore the data and try to put it uniformly for a better understanding. Which stage of AI Project cycle is
spoken about?
Ans. Data Exploration stage.
4. Imagine you are building a website for a local bakery. Instead of designing the entire website at once, you
break it down into separate parts: one for the menu, one for online orders, and another for customer
reviews, making it easier to develop and update each part independently. Name the term used for breaking
the process into smaller parts.
Ans. Modularity.
5. A team is trying to understand why their product sales have declined. They brainstorm various factors that
might have contributed to this issue. What diagram would best visualize the cause and effect relationships
between the data features?

AI REFLECTION, PROJECT CYCLE AND ETHICS 35

Ans. System Map.
6. These charts show proportions and percentages between categories, by dividing a circle into proportional
segments. Each arc length represents a proportion of each category, while the full circle represents the total
sum of all the data, equal to 100%. Identify the type of chart.
Ans. Pie chart.
7. A dataset containing continuous features and corresponding house prices (labels) is given to you. You are
asked to predict the sale price of a house based on this dataset. Which machine learning algorithm would be
most suitable for this house price prediction problem?
Ans. Regression.
8. A computer vision model is tasked with inspecting apples on a conveyor belt and classifying them as 'fresh' or
'rotten'. Name the evaluation terminology used when the model incorrectly classifies a rotten apple as fresh.
Ans. False Negative.
9. An AI-powered diagnostic tool is trained on a diverse dataset to ensure accurate and unbiased predictions of
heart disease risk across different populations. Which AI principle does this exemplify?
Ans. Inclusion.
10. Rakshit and Aman are discussing features for a new mobile phone. Aman subsequently starts receiving
notifications for mobile models matching their desired specifications. Which ethical concern does this
situation highlight? [CBSE]
Ans. Data Privacy.

Assertion – Reasoning Questions:

1. Assertion (A): Computer Vision is an AI domain that works with text and speech.
Reason (R): AI technology enables machines to understand and interpret visual information.
(a) Both (A) and (R) are correct and (R) is the correct explanation of (A).
(b) Both (A) and (R) are correct but (R) is not the correct explanation of (A).
(c) (A) is correct but (R) is incorrect.
(d) (A) is incorrect but (R) is correct.
2. Assertion (A): Natural Language Understanding (NLU) focuses on text generation.
Reason (R): NLU involves constructing coherent and contextually appropriate responses.
(a) Both (A) and (R) are correct and (R) is the correct explanation of (A).
(b) Both (A) and (R) are correct but (R) is not the correct explanation of (A).
(c) (A) is correct but (R) is incorrect.
(d) (A) is incorrect but (R) is correct.
3. Assertion (A): The ‘Why’ block is the only block of 4Ws Problem canvas that speaks about the solution to
the problem.
Reason (R): The ‘Who’ block speaks about the benefits which the stakeholders would get from the solution
and how would it benefit them as well as the society.
(a) Both (A) and (R) are correct and (R) is the correct explanation of (A).
(b) Both (A) and (R) are correct but (R) is not the correct explanation of (A).
(c) (A) is correct but (R) is incorrect.
(d) (A) is incorrect but (R) is correct.
4. Assertion (A): Open-sourced websites are used for data collection.
Reason (R): Data is available free of cost in open-sourced websites.
(a) Both (A) and (R) are correct and (R) is the correct explanation of (A).

36 AI REFLECTION, PROJECT CYCLE AND ETHICS

 (b) Both (A) and (R) are correct but (R) is not the correct explanation of (A).
(c) (A) is correct but (R) is incorrect.
(d) (A) is incorrect but (R) is correct.
5. Assertion (A): The data exploration stage deals with visualisation of data.
Reason (R): Visualising data helps to quickly get a sense of the trends, relationships and patterns contained
within the Data.
(a) Both (A) and (R) are correct and (R) is the correct explanation of (A).
(b) Both (A) and (R) are correct but (R) is not the correct explanation of (A).
(c) (A) is correct but (R) is incorrect.
(d) (A) is incorrect but (R) is correct.
6. Assertion (A): A drawback/feature for learning based approach is that the learning is static.
Reason (R): A learning based approach is one in which the machine is fed with data and the desired output,
to which the machine designs its own algorithm.
(a) Both (A) and (R) are correct and (R) is the correct explanation of (A).
(b) Both (A) and (R) are correct but (R) is not the correct explanation of (A).
(c) (A) is correct but (R) is incorrect.
(d) (A) is incorrect but (R) is correct.
7. Assertion (A): AI modeling is developing algorithms which can be trained to get intelligent output.
Reason (R): Training an AI model is required so that it can understand the various pattern, rules and
features.
(a) Both (A) and (R) are correct and (R) is the correct explanation of (A).
(b) Both (A) and (R) are correct but (R) is not the correct explanation of (A).
(c) (A) is correct but (R) is incorrect.
(d) (A) is incorrect but (R) is correct.
8. Assertion (A): Each AI problem can be mapped to the AI project cycle.
Reason (R): AI Ethics principles help us guide to create better and safer AI solutions.
(a) Both (A) and (R) are correct and (R) is the correct explanation of (A).
(b) Both (A) and (R) are correct but (R) is not the correct explanation of (A).
(c) (A) is correct but (R) is incorrect.
(d) (A) is incorrect but (R) is correct.
9. Assertion (A): AI can only function with large amounts of data.
Reason(R): AI algorithms learn from patterns identified within data.
(a) Both (A) and (R) are correct and (R) is the correct explanation of (A).
(b) Both (A) and (R) are correct but (R) is not the correct explanation of (A).
(c) (A) is correct but (R) is incorrect.
(d) (A) is incorrect but (R) is correct.
10. Assertion (A): Anyone can kick Artificial Intelligent machine.
Reason(R): They have no pain receptors.
(a) Both (A) and (R) are correct and (R) is the correct explanation of (A).
(b) Both (A) and (R) are correct but (R) is not the correct explanation of (A).
(c) (A) is correct but (R) is incorrect.
(d) (A) is incorrect but (R) is correct.

AI REFLECTION, PROJECT CYCLE AND ETHICS 37

 11. Assertion (A): AI systems in banking are used to understand customer behaviour, customer loyalty, and
service quality, which then helps in predicting future trends.
Reason(R): Voice recognition systems such as Google Assistant, Amazon’s Alexa, Apple’s Siri respond to
the queries of customers.
(a) Both (A) and (R) are correct and (R) is the correct explanation of (A).
(b) Both (A) and (R) are correct but (R) is not the correct explanation of (A).
(c) (A) is correct but (R) is incorrect.
(d) (A) is incorrect but (R) is correct.
Ans. 1. (d), 2. (d), 3. (c), 4. (a), 5. (a), 6. (d), 7. (a), 8. (b), 9. (a), 10. (d), 11. (b).
Very Short Answer Type Questions
1. Name the term that refers to the capability of machines to perform cognitive tasks such as thinking,
perceiving, learning, problem-solving, and decision-making.
Ans. Artificial Intelligence.
2. Give two examples of e-commerce platforms.
Ans. Flipkart, Amazon.
3. What type of input data is fed into data domain?
Ans. Numerical data and Alphanumerical data
4. Name the fourth stage of AI Project Cycle.
Ans. Modelling.
5. Name the framework that helps you in identifying the key elements related to a problem.
Ans. 4Ws Problem Canvas.
6. Give two examples of data sources.
Ans. Surveys and open-source websites.
7. List any five graphs used for data visualization. [CBSE]
Ans. Bar chart, Pie chart, Scatter chart, Line chart, Histogram.
8. Name the type of machine learning where a model is trained on labeled data.
Ans. Supervised learning.
9. Give two examples of platforms on which an AI project can be deployed.
Ans. Mobile Applications and Web Applications.
10. Identify whether the following question is related to ethics or morals:
Is taking pens from a library considered stealing?
Ans. Ethics.
11. Name two principles in AI Ethics affect the quality of AI solutions.
Ans. Human Rights, Bias, Privacy, Inclusion.
Short Answer Type Questions
1. Define Artificial Intelligence.
Ans. Artificial intelligence (AI) may be defined as a machine’s capability to recognize patterns and make
predictions based on them.
2. Name any five applications of AI systems.
Ans. Education & Training, Customers Sport Systems, Service Oriented Business, Product Oriented Business, E-
commerce & Retail Business
3. Enlist the three Domains of AI.

38 AI REFLECTION, PROJECT CYCLE AND ETHICS

 Ans. Data. Computer Vision and Natural Language Processing.
4. How is AI helpful in service oriented business?
Ans. In service-oriented businesses like banking, education, travel, the main asset is data. AI systems accept the
data, works on it, and help its users in bringing out positive results for the business. These systems are also
used to understand customer behaviour, customer loyalty, and service quality, which then helps in
predicting future trends and lot more.
5. How many domains are there in AI? Name them.
Ans. AI is composed of the following three domains—
(a) Data
(b) Computer Vision
(c) Natural Language Processing (NLP)
6. How is an Al project different from an IT project? [CBSE]
Ans. AI projects focus on creating intelligent systems that learn and adapt with data, while IT projects primarily
manage and implement existing systems.
7. Name the 4Ws of the 4Ws Problem Canvas.
Ans. Who, What, Where and Why.
8. Why is there a need to use a Problem Statement Template during problem scoping? [CBSE]
Ans. A problem statement template is a structured framework that summarises the essence of a problem,
including its impact, context, and potential solutions. It serves as a foundation for further analysis and
development.
9. Differentiate between Training Data and Testing Data. [CBSE]
Ans. Training data Testing data
Is the one on which we train and fit Is used only to assess performance of model
our model basically to fit the parameters
Output is available to model The unseen data for which predictions have to be made

10. Name a few government websites from where you can get open-source data. [CBSE]
Ans. data.gov.in, india.gov.in
11. Why we need to explore data?
Ans. We need to explore data to
• Get a sense of trends, relationships and patterns present in data
• Decide which model to use in our AI Project cycle
• Easier to comprehend information
• Easier to communicate the story with others
12. How is Machine Learning related to Artificial Intelligence? [CBSE]
Ans. Machine Learning (ML) is a subset of AI that specifically focuses on enabling machines to learn from data
which may or may not involve explicit programming.
13. What is Evaluation? [CBSE]
Ans. Evaluation is the process of understanding the reliability of any AI model, based on outputs by feeding test
dataset into the model and comparing with actual answers.
14. What are various Model evaluation techniques? [CBSE]
Ans. TP (True Positive): The model correctly predicted a positive class.

AI REFLECTION, PROJECT CYCLE AND ETHICS 39

 FP (False Positive): The model incorrectly predicted a positive class when it was actually negative.
TN (True Negative): The model correctly predicted a negative class.
FN (False Negative): The model incorrectly predicted a negative class when it was actually positive.
15. Why is model evaluation important in AI projects? [CBSE]
Ans. Model evaluation ensures that the developed model is reliable, accurate, and fit for its intended purpose.
16. What do you understand by the terms True Positive and False Positive? [CBSE]
Ans. TP (True Positive): The model correctly predicted a positive class.
FP (False Positive): The model incorrectly predicted a positive class when it was actually negative.
17. Define principles of AI. [CBSE]
Ans. Principles of AI are a set of guidelines that govern the design, development, and deployment of artificial
intelligence systems. These principles aim to ensure that AI is developed and used ethically, responsibly, and
beneficially for society.
18. Write Major Issues around AI Ethics. [CBSE]
Ans. • Bias: This encompasses algorithmic bias, which can lead to discriminatory outcomes in various fields.
• Privacy: The collection and use of personal data for AI development raises serious privacy concerns.
• Job Displacement: Automation driven by AI has the potential to significantly impact the job market.

Long Answer Questions

1. Explain the 4Ws problem canvas in problem scoping.
Ans. Who
This block is used to identify the stakeholders, or those who are directly or indirectly affected by the problem.
It is important to understand their needs, perspectives, and priorities. This research provides valuable
insights into the problem’s impact on different groups.
What
This block is used to clearly define the problem you are addressing. Articulate the issue at hand, including its
scope and severity, to establish a clear focus. Gathering evidence such as data, statistics, and case studies will
support the existence of the problem and make it more credible.
Where
This block is used to understand the context in which the problem occurs by analyzing the environment. This
helps in identifying the geographical distribution of the problem, pinpointing specific locations where it is
most prevalent. This knowledge can aid in creating targeted solutions.
Why
This block is used to assess the impact of the problem by evaluating its consequences on individuals and
society. Define the potential benefits of a solution to motivate action and garner support. This involves
articulating the value of solving the problem and how it will positively affect the stakeholders involved.
2. Draw the problem statement template.
Ans. Our [stakeholders] Who?
have/has a problem that [issue/problem/need] What?
When/while [context/situation]. Where?
An ideal solution would be [benefit of solution]. Why?

40 AI REFLECTION, PROJECT CYCLE AND ETHICS

 3. What is a system map? How do you draw it? [CBSE]
Ans. A system map is a visual representation that shows the components and boundaries of a system at a specific
point in time. It helps us understand the relationships between these components and how they interact with
each other to produce a specific outcome. The arrows in a system map depict the cause-and-effect
relationships between components.
To draw the system map,
• Determine the data features that influence the system.
• Represent each data feature with a circle and label each circle with the name.
• Indicate the relationships between the data features by drawing arrows pointing from one factor to
another to show the direction of influence.
• A positive (+) sign indicates a positive correlation or direct relationship: as one factor increases, the
other tends to increase as well.
• A negative (-) sign indicates a negative correlation or indirect relationship: as one factor increases, the
other tends to decrease.
4. Explain any one Data Visualization technique.
Ans.

5. Identify A, B and C in the following diagram (Hint: How AI, ML &DL related to each other) [CBSE]

c
b
a

AI REFLECTION, PROJECT CYCLE AND ETHICS 41

 (a) Artificial Intelligence (AI) is the broader field that encompasses the development of intelligent
agents, which are systems that can reason, learn, and act autonomously.
(b) Machine Learning (ML) is a subset of AI that focuses on enabling systems to learn from data without
being explicitly programmed. ML algorithms identify patterns in data and use them to make predictions
or decisions.
(c) Deep Learning (DL) is a specialized subset of machine learning that utilizes artificial neural networks
with multiple layers to analyze complex patterns in large datasets. Deep learning has been instrumental
in advancements such as image recognition, natural language processing, and speech recognition.
6. Compare and contrast Rule-based and Learning-based approach in Al modeling indicating clearly when
each of these may be used. [CBSE]
Ans. (a) Traditional Rule-Based Systems operate on a set of predefined rules and logic, similar to following a
detailed instruction manual. When these systems encounter a problem, they consult their established
rules to determine the appropriate decision.
• One of the main strengths of rule-based systems is their transparency and explainability; the
decision-making process is clear and straightforward, making it easy to understand how conclusions
are reached. Additionally, these systems are highly efficient in domains where rules are well-defined
and consistent.
• However, rule-based systems also have notable limitations. They are rigid and inflexible, struggling
to adapt to new or unexpected situations. This rigidity makes them less effective when dealing with
complex problems involving multiple variables.
(b) Learning based systems learn from data and improve their performance over time. They identify
patterns and relationships within data to make predictions or decisions. They can be further divided into
two main types: machine learning and deep learning. Machine learning systems learn from data
without needing explicit programming for each task, while deep learning, a subset of machine learning,
utilizes artificial neural networks with multiple layers to analyze complex patterns in data.
The strengths of learning-based systems include –
• Adaptability: Can handle complex and evolving problems.
• High accuracy in pattern recognition tasks.
• Continuous improvement through learning.
However, they also come with limitations. Learning-based systems often require large amounts of data to
function effectively and can be computationally expensive due to the resources needed for processing and learning.
Furthermore, these systems can lack transparency in their decision-making processes, making it harder to
understand how decisions are reached.
7. Differentiate between Ethics and Moral with suitable examples. [CBSE]
Ans.

Morals Morals

Morals primarily focus on the individual. Ethics are concerned with societal norms.

They are derived from personal beliefs, upbringing, They stem from shared values and community
and culture. agreements.

42 AI REFLECTION, PROJECT CYCLE AND ETHICS

 They tend to be subjective, reecting personal They are considered more objective, based on
perspectives. broader societal standards.

They function as an internal compass guiding They provide an external framework for conduct.
individual behaviour.

Examples of morals include honesty, fairness, and Examples of ethics encompass professional codes,
compassion. laws, and human rights.

8. A company had been working on a secret AI recruiting tool. The machine-learning specialists uncovered a
big problem: their new recruiting engine did not like women chefs. The system taught itself that male
candidates are preferable. It penalised resumes that included the word “women chef". This led to the failure
of the tool.
(a) What aspect of AI ethics is illustrated in the given scenario?
(b) What could be the possible reasons for the ethical concern identified? [CBSE]
Ans. (a) Bias is illustrated in the given scenario.
(b) • The dataset used to train the AI might have contained historical biases, reflecting societal biases
where male chefs were more common or preferred. This bias in the data would have been learned
and replicated by the AI model.
• If the training data lacked sufficient representation of female chefs or women in similar roles, the AI
system would not have learned to value or recognize female candidates appropriately.

Check Your Progress

1. Who coined the term :Artificial Intelligence:?

(a) Charles Babbage (b) John McCarthy
(c) Dave Waters (d) Stepehen Hawking
2. The Third of project cycle is ................... .
(a) Data Acquisition (b) Data Exploration
(c) Evaluation (d) Deployment
3. Arrange the following in correct order:
(i) Identify a theme for the problem (ii) Identify the subtopic
(iii) Choosing a topic (iv) Identify the problem
(a) ii, i, iii, iv (b) i, ii, iv, iii
(c) iv, i, iii, ii (d) i, iii, ii, iv
4. “Preference for one over the other” is known as ____________.
5. __________ as the last stage in the AI project cycle where the AI model or solution is implemented in a real-
world scenario.
6. When prediction = Yes, and Reality = Yes, the model terminology used is False Positive. (True/False)
7. In a system map, if an arrow goes from X to Y with a – (minus) sign, it means that if X increases, Y decreases.
(True/False)

AI REFLECTION, PROJECT CYCLE AND ETHICS 43

 8. Assertion (A): Open-sourced websites are used for data collection.
Reason (R): Data is available free of cost in open-sourced websites.
(a) Both (A) and (R) are correct and (R) is the correct explanation of (A).
(b) Both (A) and (R) are correct but (R) is not the correct explanation of (A).
(c) (A) is correct but (R) is incorrect.
(d) (A) is incorrect but (R) is correct.
9. Assertion (A): Each AI problem can be mapped to the AI project cycle.
Reason (R): AI Ethics principles help us guide to create better and safer AI solutions.
(a) Both (A) and (R) are correct and (R) is the correct explanation of (A).
(b) Both (A) and (R) are correct but (R) is not the correct explanation of (A).
(c) (A) is correct but (R) is incorrect.
(d) (A) is incorrect but (R) is correct.
10. What type of input data is fed into data domain?
11. Name the type of machine learning where a model is trained on labeled data.
12. A dataset containing continuous features and corresponding house prices (labels) is given to you. You are
asked to predict the sale price of a house based on this dataset. Which machine learning algorithm would be
most suitable for this house price prediction problem?
13. An AI-powered diagnostic tool is trained on a diverse dataset to ensure accurate and unbiased predictions of
heart disease risk across different populations. Which AI principle does this exemplify?
14. Define the term “Artificial Intelligence”.
15. Name any two data visualisation tools.
16. Differentiate between training data and testing data.
17. How is Deep Learning related to Artificial Intelligence?
18. Explain any two AI Principles that guide the development and deployment of responsible AI.

Lab Activities

1. Microsoft Copilot is a tool that uses generative AI to serve as a helpful assistant in the field of
education. Give examples of ways in which Microsoft Copilot can be used.
Sol. • Personalized learning: Copilot can support personalized learning by helping you create content,
tailored feedback, and guidance for students based on their individual needs and learning styles.
• Brainstorming: You can use Copilot to brainstorm new ideas for activities, lesson plans, supporting
materials, and assignments.
• Lesson planning: Copilot can help you plan lessons by suggesting or drafting activities, resources, and
assessments that align with learning objectives. You can also use Copilot to start a rubric for the lessons.
• Provide feedback: Copilot can help you draft initial feedback and ideas for students on their work,
which you can edit and personalize for your students.
• Get quick answers: Copilot can help you get quick answers to your questions without having to read
through multiple search results. Also, Copilot provides links to content sources so you can assess the
source or dive deeper into the original content.

44 AI REFLECTION, PROJECT CYCLE AND ETHICS

 2. Explain how the game ‘Rock, Paper, Scissors’ work after playing it from the website
https://next.rockpaperscissors.ai/
Sol. About this Activity. Rock paper scissors is an old and
simple game played with the hands between two players. It
can also now be played with system as one player and you
will be the second player.
Method. It has only three rules–
1. Paper beats rock.
2. Scissors beats paper.
3. Rock beats scissors.
The players of the game have to guess what the opponents
has in mind and make an appropriate shape out of the
three that is rock/paper/scissors.
AI Working. The AI program keeps track of the choices made by you and assigns probability to it. Based on
the probability, it responds with appropriate shapes and will try to defeat you. For this AI system uses
Machine learning domain.
3. What do you infer after playing the NLP based game Semantris
(https://experiments.withgoogle.com/semantris)?
Ans. Semantris is a word association game powered by Artificial Intelligence (AI). It presents you with a keyword
and challenges you to provide words related to it. The game has two modes: "Arcade" for a fast-paced
challenge and "Blocks" for a more relaxed experience.
• AI can grasp word relationships: The game uses semantic search, suggesting the AI can
understand connections between words based on meaning, context, and even synonyms/antonyms.
• Focus on natural conversation: My experience indicates the AI excels at recognizing common word
associations used in everyday talk.
• Room for improvement: There were some unexpected results, implying the AI might struggle with
less common or creative connections.

4. Write down the steps to play the Computer Vision based application Autodraw
https://www.autodraw.com/?
Ans. • Navigate to experiments.withgoogle.com/autodraw using your preferred web browser.
• Select the "AutoDraw" tool and commence free hand drawing. Use either your mouse or touchpad as
your creative instrument, focusing on expressing your artistic vision without concern for technical
perfection.
• As your drawing progresses, AutoDraw actively analyzes the lines and shapes you create. It aims to

AI REFLECTION, PROJECT CYCLE AND ETHICS 45

 decipher your artistic intent and the image you are
forming.
• Based on its analysis, it curates and presents a
selection of relevant images from its extensive library.
These suggestions may encompass basic shapes or
even more intricate illustrations.
• Select the image that aligns most closely with your
artistic vision or continue drawing to further refine
your creation.
• Once done, download it as a file for seamless integration into a worksheet or save it as an image file for
future reference.
5. The COVID-19 pandemic posed unprecedented challenges to global health systems. Rapidly
increasing infection rates overwhelmed healthcare infrastructure, making it difficult to
effectively track and contain the virus's spread. Traditional contact tracing methods, reliant on
manual data collection and analysis, proved inefficient in the face of the pandemic's scale. This
led to delays in identifying potential exposure points, isolating infected individuals, and
implementing containment measures. Consequently, the virus continued to spread rapidly,
resulting in high infection and mortality rates. A robust contact tracing system is essential for
breaking the chain of infection and controlling disease outbreaks.
Sol. Frame a problem statement template for the given scenario.
Our citizens Who?
have/has a problem that there is a difficulty in effectively tracking and What?
containing the spread of COVID-19
When/while a global pandemic emerged, overwhelming Where?
healthcare systems and leading to rapid disease
transmission.
An ideal solution would be a robust contact tracing system that Why?
can efficiently identify and isolate infected
individuals to prevent further spread.

6. Create a system map illustrating the relationship between automation, profits, job loss, stress, and
frustration.
Sol.

46 AI REFLECTION, PROJECT CYCLE AND ETHICS

 7. Livegap is a portal to create different types of charts based on the available data. Demonstrate
the use of this portal for visualizing data.
Sol. (i) Open the Livegap site in your borwser using the URL given.
(ii) Click on the Livegap Chart to begin with the activity.

(iii) On the next screen click on Make your Chart button.

(iv) Select a particular type of chart you want to create.

(v) Enter data in the table given for creating a chart,

(vi) You can work with various components of the chart using the Sidebar.
(vii) Once you are done you can also save and share the chart created.

AI REFLECTION, PROJECT CYCLE AND ETHICS 47

 8. Use teachable machine to demonstrate how an AI model can be evaluated.
Ans (i) Go to the website https://teachablemachine.withgoogle.com/

(ii) Click on Get Started.

(iii) Teachable machine shows three options.

(iv) Select the ‘Image’ project.

(v) Select ‘Standard Image Model’.

48 AI REFLECTION, PROJECT CYCLE AND ETHICS

 (vi) You will get a screen like this.

(vii) Download pictures (a sample of 5 each) of oranges and lemons in another tab.
(viii)Select upload option, upload pictures of Oranges in Class 1 and rename it as “Oranges”
(ix) Next, Select upload option, upload pictures of Lemons in Class 2 and rename it as “Lemons”
(x) Click on Train Model.
(xi) Once the model is trained, change the Input from Webcam to File and upload a sample picture that is
distinct from what was uploaded under Oranges/Lemons earlier.
(xii) You will find that the model is predicting the given picture under one of the categories.

AI REFLECTION, PROJECT CYCLE AND ETHICS 49

 9. An e-commerce platform is experiencing a surge in fraudulent transactions, resulting in
significant financial losses. The company aims to build a robust fraud detection system to
identify and prevent fraudulent activities.
Sol.

Data Data
Modelling Evaluation Deployment
Acquisition Exploration

Collect a Analyze the Develop and train Evaluate model Integrate the
dataset dataset to machine learning performance using model into a
containing understand models (e.g., metrics like healthcare
patient records correlations logistic regression, accuracy, system or
with attributes between obesity decision trees, precision, recall, develop a user-
like, age, and different random forest) and F1-score. friendly
gender, height, factors. Identify to predict obesity Fine-tune the application for
weight, potential features risk based on model based on obesity risk
blood pressure, for model the collected data. evaluation assessment.
cholesterol training. Handle results. Provide
levels, and missing values recommendations
dietary habits, and outliers. for users based
on the
prediction.

10. Use the Moral Machine platform to analyze the ethical dilemmas faced by autonomous
vehicles. What factors influence your decision-making when confronted with life-or-death
choices? How do your choices compare to societal norms and preferences as reflected in the
platform's data?
Sol. The Moral Machine presents users with hypothetical scenarios where autonomous vehicles must make split-
second decisions that could impact human lives. These scenarios often involve moral conflicts, such as
deciding whether to prioritize the safety of passengers or pedestrians, elderly individuals versus children, or
obeying traffic laws versus avoiding greater harm.

50 AI REFLECTION, PROJECT CYCLE AND ETHICS

 The decisions uncover whether:
• Saving lives is paramount, regardless of the circumstances.
• Legal considerations influence your choices when lives are at stake.
• Protecting specific groups takes precedence over others.

AI REFLECTION, PROJECT CYCLE AND ETHICS 51