Artificial Intelligence (AI)

Artificial Intelligence (AI) is the branch of computer science that enables machines and computer systems to perform tasks that normally require human intelligence.

These tasks include:

Learning
Reasoning
Problem-solving
Decision-making
Understanding language
Recognising images and speech
Perception
Planning

Simple Definition

AI is the simulation of human intelligence in machines that are programmed to think, learn and act like humans.

Important Definitions

Term	Definition
Artificial Intelligence	Simulation of human intelligence by machines
Machine Intelligence	Ability of a machine to perform intelligent tasks
Machine Learning	A method through which machines learn from data
Deep Learning	A subset of ML using multi-layered neural networks
Neural Network	Computing system inspired by the human brain
Robotics	Design and use of programmable machines/robots
Natural Language Processing	AI technology that enables computers to understand human language
Computer Vision	AI technology that enables computers to interpret images and videos
Expert System	AI system that imitates the decision-making ability of a human expert

History of Artificial Intelligence

Important Events

Year	Event
1943	Warren McCulloch and Walter Pitts proposed a mathematical model of artificial neurons
1950	Alan Turing proposed the Turing Test
1956	The term Artificial Intelligence was coined at the Dartmouth Conference
1956	John McCarthy is widely regarded as the Father of Artificial Intelligence
1960s	Development of early AI programs and expert systems began
1966	ELIZA, an early chatbot, was developed by Joseph Weizenbaum
1970s–1980s	Expert systems became popular
1997	IBM’s Deep Blue defeated chess champion Garry Kasparov
2011	IBM Watson defeated human champions on the quiz show Jeopardy!
2016	Google’s AlphaGo defeated Go champion Lee Sedol
2022 onwards	Generative AI and large language models became widely popular

Important Personalities

John McCarthy – Father of Artificial Intelligence
Alan Turing – Pioneer of Computer Science and AI
Marvin Minsky – Pioneer of AI research
Herbert Simon – Pioneer of Artificial Intelligence
Allen Newell – AI researcher
Geoffrey Hinton – Known as one of the pioneers of modern deep learning
Yann LeCun – Pioneer of convolutional neural networks
Yoshua Bengio – Pioneer of deep learning

The Turing Test

The Turing Test was proposed by Alan Turing in 1950.

Purpose

To determine whether a machine can demonstrate intelligent behaviour similar to a human.

Basic Idea

A human evaluator communicates with:

A human
A machine

If the evaluator cannot reliably distinguish between them, the machine is considered to have passed the test.

MCQ Point

The Turing Test was proposed by Alan Turing.

Types of Artificial Intelligence

AI can be classified in different ways.

A. Based on Capability

1. Narrow AI / Weak AI

Designed to perform a specific task.

Examples:

Voice assistants
Face recognition
Recommendation systems
Spam filters
Chess-playing programs
Chatbots

Most AI systems used today are Narrow AI.

2. General AI / Strong AI

A hypothetical AI system capable of performing any intellectual task that a human can perform.

It would be able to:

Learn different subjects
Reason
Adapt
Solve new problems

General AI has not yet been fully achieved.

3. Super AI

A hypothetical form of AI that would surpass human intelligence in almost every field.

It is currently theoretical.

B. Based on Functionality

1. Reactive Machines

Simplest type of AI
Cannot store memories
Respond only to current situations

Example:

IBM Deep Blue

2. Limited Memory AI

Can use past data for a limited period.

Examples:

Self-driving vehicles
Many modern AI systems
Recommendation systems

3. Theory of Mind AI

A theoretical AI capable of understanding:

Human emotions
Beliefs
Intentions
Thoughts

4. Self-Aware AI

A hypothetical AI possessing consciousness and self-awareness.

Machine Learning (ML)

Machine Learning is a subset of Artificial Intelligence.

Definition

Machine Learning enables computers to learn from data and improve their performance without being explicitly programmed for every task.

Traditional Programming

Rules + Data → Output

Machine Learning

Data + Expected Output → Rules/Model

Types of Machine Learning

1. Supervised Learning

The machine learns from labelled data.

Example:

Input	Correct Answer
Picture of a cat	Cat
Picture of a dog	Dog

The model learns the relationship between input and output.

Applications:

Classification
Prediction
Spam detection
Disease prediction
House price prediction

Common Algorithms:

Linear Regression
Logistic Regression
Decision Tree
Random Forest
Support Vector Machine
K-Nearest Neighbours

2. Unsupervised Learning

The machine learns from unlabelled data.

It identifies patterns and groups within the data.

Applications:

Customer segmentation
Pattern recognition
Market analysis

Common Algorithms:

K-Means Clustering
Hierarchical Clustering
Principal Component Analysis (PCA)

3. Semi-Supervised Learning

Uses a combination of:

A small amount of labelled data
A large amount of unlabelled data

Reinforcement Learning

The machine learns through:

Actions
Rewards
Penalties

Basic Concept

Agent → Action → Environment → Reward/Penalty

Examples:

Game-playing AI
Robotics
Autonomous vehicles

Important Terms

Term	Meaning
Agent	Learner or decision-maker
Environment	Surrounding world
Action	Decision taken by agent
Reward	Positive feedback
Penalty	Negative feedback
Policy	Strategy used by the agent

Deep Learning

Deep Learning is a subset of Machine Learning.

It uses artificial neural networks with multiple layers.

Structure

Input Layer → Hidden Layers → Output Layer

Deep learning is especially useful for:

Image recognition
Speech recognition
Natural Language Processing
Autonomous vehicles
Medical imaging

Common Deep Learning Networks

Network	Main Use
CNN	Images and computer vision
RNN	Sequential data
LSTM	Long-term sequential data
GAN	Generating new content
Transformer	Language and generative AI

Artificial Neural Network (ANN)

An Artificial Neural Network is inspired by the structure of the human brain.

Basic Unit: Artificial Neuron

An artificial neuron receives:

Inputs
Weights
Bias

It processes them and produces an output.

Main Components

Input Layer
Hidden Layer
Output Layer

Important Terms

Neuron – Basic processing unit
Weight – Importance assigned to an input
Bias – Additional parameter
Activation Function – Determines output of a neuron
Epoch – One complete pass through training data
Batch – Group of training examples

Natural Language Processing (NLP)

NLP enables computers to understand, interpret and generate human language.

Applications

Chatbots
Language translation
Voice assistants
Sentiment analysis
Speech recognition
Text summarisation
Grammar checking

Examples:

Google Translate
Chatbots
Voice assistants
Text prediction systems

Important NLP Terms

Tokenisation

Breaking text into smaller units called tokens.

Example:

Artificial Intelligence is useful.

Tokens:

Artificial | Intelligence | is | useful

Sentiment Analysis

Determines whether a text expresses:

Positive sentiment
Negative sentiment
Neutral sentiment

Speech Recognition

Converts:

Speech → Text

Text-to-Speech (TTS)

Converts:

Text → Speech

Machine Translation

Automatically translates text from one language to another.

Generative Artificial Intelligence

Generative AI is AI capable of creating new content.

It can generate:

Text
Images
Audio
Video
Code
Music

Examples:

AI chatbots
Text-to-image systems
AI music generators
AI coding assistants

Generative AI vs Traditional AI

Traditional AI	Generative AI
Analyses data	Creates new content
Makes predictions	Generates text, images, audio, etc.
Classifies information	Produces original outputs

Large Language Model (LLM)

A Large Language Model is an AI model trained on huge amounts of text data.

It can:

Understand language
Generate text
Answer questions
Summarise documents
Translate languages
Write code

Examples of LLM-Based Systems

Chatbots
AI writing assistants
Coding assistants

Transformer Technology

The Transformer architecture is a deep learning architecture widely used in modern NLP and Generative AI.

Key Feature

Attention mechanism

The attention mechanism helps a model understand the relationship between different words in a sentence.

Important Concept

Transformers use attention mechanisms to process relationships between words and tokens.

Large Language Model (LLM) Important Terms

Training

Process of teaching a model using large amounts of data.

Inference

The process of using a trained AI model to generate an answer or prediction.

Prompt

Instruction or input given to an AI system.

Prompt Engineering

Designing effective prompts to obtain better results from an AI model.

Hallucination

When an AI system generates information that is false, inaccurate or unsupported but appears convincing.

Fine-Tuning

Further training a pre-trained model for a specific task or domain.

Pre-trained Model

A model already trained on a large dataset.

Computer Vision

Computer Vision enables computers to understand and interpret visual information.

Applications:

Face recognition
Object detection
Medical image analysis
Self-driving cars
Security systems
OCR
Image classification

Important Terms

Image Classification

Identifying what is present in an image.

Object Detection

Identifying objects and their location in an image.

Facial Recognition

Identifying or verifying a person based on facial features.

OCR

Optical Character Recognition converts images of text into machine-readable text.

Robotics and AI

Robotics involves the design, construction and operation of robots.

AI enables robots to:

Perceive their surroundings
Make decisions
Learn
Navigate
Perform tasks

Examples:

Industrial robots
Medical robots
Humanoid robots
Autonomous robots
Drones

Important Difference

AI	Robotics
Software intelligence	Physical machine
Makes decisions	Performs physical actions
Can exist without a robot	Often uses AI for intelligence

Expert Systems

An Expert System is an AI system that imitates the decision-making ability of a human expert.

Main Components

Knowledge Base
Inference Engine
User Interface

Knowledge Base

Stores facts and rules.

Inference Engine

Applies rules to facts and produces conclusions.

Expert System Structure

User → User Interface → Inference Engine → Knowledge Base

AI and Big Data

AI requires large amounts of data for:

Training
Pattern recognition
Prediction
Decision-making

Relationship

Big Data provides data, while AI analyses that data to generate insights.

Important AI Technologies

Technology	Function
Machine Learning	Learns from data
Deep Learning	Uses multi-layer neural networks
NLP	Understands human language
Computer Vision	Understands images and videos
Robotics	Builds intelligent machines
Expert Systems	Imitates expert decision-making
Speech Recognition	Converts speech into text
Generative AI	Creates new content
Reinforcement Learning	Learns through rewards and penalties
Fuzzy Logic	Handles uncertainty and approximate reasoning

Fuzzy Logic

Traditional logic generally uses:

True or False
0 or 1

Fuzzy logic allows degrees of truth.

Example:

A temperature may be:

Slightly hot
Moderately hot
Very hot

Applications:

Washing machines
Air conditioners
Control systems
Automotive systems

AI in Different Fields

Healthcare

AI is used for:

Disease detection
Medical imaging
Drug discovery
Patient monitoring
Virtual health assistants

Agriculture

AI helps in:

Crop monitoring
Weather prediction
Pest detection
Soil analysis
Smart irrigation

Education

AI is used for:

Personalised learning
Automated evaluation
Intelligent tutoring systems
AI-based educational assistants

Banking and Finance

AI is used for:

Fraud detection
Credit scoring
Algorithmic trading
Customer service chatbots
Risk assessment

Transportation

AI is used in:

Self-driving vehicles
Traffic management
Route optimisation
Predictive maintenance

Defence

AI is used for:

Surveillance
Cybersecurity
Intelligence analysis
Autonomous systems

Governance

AI can assist in:

Public service delivery
Fraud detection
Disaster management
Policy analysis
Citizen services

Advantages of Artificial Intelligence

Reduces human error
Works continuously
Performs repetitive tasks
Processes large amounts of data
Provides fast decision-making
Improves efficiency
Can work in dangerous environments
Supports scientific research

Limitations of AI

High development cost
Dependence on data
Lack of human emotions
Potential bias
Privacy concerns
Job displacement
Security risks
Lack of common sense
AI hallucinations
Dependence on computing resources

AI Bias

AI bias occurs when an AI system produces unfair or prejudiced results.

Causes:

Biased training data
Incomplete data
Poor algorithm design
Human bias

Example:

If an AI recruitment system is trained on historically biased hiring data, it may produce biased hiring recommendations.

AI Ethics

Important ethical issues include:

Privacy
Transparency
Accountability
Fairness
Safety
Human control
Data protection

Important Principle

AI systems should be developed and used responsibly, fairly and transparently.

AI and Privacy

AI systems often require large amounts of personal data.

Potential risks include:

Unauthorised data collection
Surveillance
Data leaks
Misuse of biometric data
Identity theft

Explainable AI (XAI)

Explainable AI refers to AI systems whose decisions can be understood and explained by humans.

Importance:

Builds trust
Improves accountability
Helps identify errors
Useful in healthcare and finance

Responsible AI

Responsible AI focuses on:

Fairness
Safety
Privacy
Transparency
Accountability
Human oversight

AI and Cybersecurity

AI for Cybersecurity

AI can detect:

Malware
Suspicious activities
Network attacks
Fraud
Unusual behaviour

AI Against Cybersecurity

AI can also be misused for:

Automated cyberattacks
Deepfakes
Phishing
Malware development
Social engineering

Deepfake Technology

A Deepfake is synthetic media created or manipulated using AI.

It can alter:

Face
Voice
Video
Audio

Risks:

Misinformation
Fraud
Identity theft
Political manipulation

AI and Cloud Computing

Cloud computing provides:

Computing power
Storage
AI services
Machine learning platforms

AI systems can be accessed through cloud platforms without owning expensive hardware.

AI and Internet of Things (IoT)

When AI is combined with IoT, it is often called:

AIoT – Artificial Intelligence of Things

Example:

Smart devices collect data through sensors and AI analyses it to make decisions.

33. Edge AI

Edge AI means processing AI data locally on a device rather than sending everything to a central cloud server.

Advantages:

Faster response
Lower latency
Better privacy
Reduced internet dependence

Examples:

Smartphones
Smart cameras
Autonomous vehicles

AI Hardware

CPU

General-purpose processor.

GPU

Highly suitable for parallel processing and AI computations.

TPU

Tensor Processing Unit, designed specifically for machine learning workloads.

NPU

Neural Processing Unit, designed for AI and neural network operations.

Important AI Terms

Term	Meaning
Algorithm	Step-by-step procedure for solving a problem
Dataset	Collection of data
Feature	Input characteristic used by a model
Label	Correct output associated with data
Model	Trained computational system
Training Data	Data used to train a model
Testing Data	Data used to evaluate a model
Validation Data	Data used to tune a model
Prediction	Output generated by a model
Accuracy	Percentage of correct predictions
Bias	Error due to overly simple assumptions
Variance	Sensitivity to training data
Overfitting	Model performs well on training data but poorly on new data
Underfitting	Model is too simple to learn the data properly

Confusion Matrix

A confusion matrix evaluates classification models.

Term	Meaning
True Positive (TP)	Correctly predicted positive
True Negative (TN)	Correctly predicted negative
False Positive (FP)	Incorrectly predicted positive
False Negative (FN)	Incorrectly predicted negative

Important AI Algorithms

Linear Regression

Used for predicting continuous values.

Logistic Regression

Used mainly for classification.

Decision Tree

Uses a tree-like decision structure.

Random Forest

Combines multiple decision trees.

K-Means

Used for clustering.

K-Nearest Neighbours

Classifies data based on nearby examples.

Support Vector Machine

Used for classification and regression.

Neural Networks

Used for complex pattern recognition.

AI, ML and Deep Learning Relationship

Artificial Intelligence
        ↓
Machine Learning
        ↓
Deep Learning

Remember:

All Deep Learning is Machine Learning, and all Machine Learning is a part of AI.

But:

Not all AI is Machine Learning.

AI vs Machine Learning vs Deep Learning

Feature	AI	ML	Deep Learning
Meaning	Broad field of intelligent machines	Learning from data	Advanced ML using neural networks
Data Requirement	Variable	Usually significant	Usually very large
Human Intervention	Can be high	Moderate	Often lower
Technology	Rules, logic, ML	Algorithms	Deep neural networks

Important AI Abbreviations

Abbreviation	Full Form
AI	Artificial Intelligence
ML	Machine Learning
DL	Deep Learning
ANN	Artificial Neural Network
CNN	Convolutional Neural Network
RNN	Recurrent Neural Network
LSTM	Long Short-Term Memory
NLP	Natural Language Processing
NLU	Natural Language Understanding
NLG	Natural Language Generation
OCR	Optical Character Recognition
ASR	Automatic Speech Recognition
TTS	Text-to-Speech
GAN	Generative Adversarial Network
LLM	Large Language Model
GPT	Generative Pre-trained Transformer
XAI	Explainable Artificial Intelligence
AGI	Artificial General Intelligence
ANI	Artificial Narrow Intelligence
ASI	Artificial Superintelligence
RL	Reinforcement Learning
IoT	Internet of Things
AIoT	Artificial Intelligence of Things
TPU	Tensor Processing Unit
GPU	Graphics Processing Unit
NPU	Neural Processing Unit
RPA	Robotic Process Automation
OCR	Optical Character Recognition
API	Application Programming Interface
GAN	Generative Adversarial Network
PCA	Principal Component Analysis
SVM	Support Vector Machine

AI in India

India has focused on the development and application of AI in areas such as:

Healthcare
Agriculture
Education
Smart cities
Governance
Financial services
Language technology
Disaster management

Important Indian AI Initiatives

IndiaAI Mission

A national initiative aimed at strengthening India’s AI ecosystem through:

AI computing capacity
AI innovation
Datasets
Startup support
AI skills
Responsible AI

Bhashini

An initiative focused on developing language technology and making digital services accessible in Indian languages.

Exam Tip

Questions on India’s AI initiatives may ask about:

AI infrastructure
Indian language technology
Digital governance
AI startups
Ethical and responsible AI

AI may:

Replace Some Tasks

Particularly:

Repetitive work
Routine data processing
Basic customer support

Create New Jobs

Such as:

AI Engineer
Data Scientist
Machine Learning Engineer
Prompt Engineer
AI Ethics Specialist
Robotics Engineer

Important Concept

AI is more likely to transform many jobs and tasks than simply eliminate all jobs.

John McCarthy is known as the Father of Artificial Intelligence.
The term Artificial Intelligence was coined in 1956.
The term AI was associated with the Dartmouth Conference.
Alan Turing proposed the Turing Test.
Machine Learning is a subset of AI.
Deep Learning is a subset of Machine Learning.
CNNs are widely used for image processing.
RNNs are useful for sequential data.
NLP deals with human language.
Computer Vision deals with images and videos.
OCR converts images of text into machine-readable text.
Reinforcement Learning uses rewards and penalties.
Supervised Learning uses labelled data.
Unsupervised Learning uses unlabelled data.
Semi-supervised Learning uses both labelled and unlabelled data.
Generative AI creates new content.
LLM stands for Large Language Model.
GPT stands for Generative Pre-trained Transformer.
GAN stands for Generative Adversarial Network.
TPU is designed for machine learning operations.
Deep Blue defeated Garry Kasparov in chess.
AlphaGo defeated Lee Sedol in Go.
ELIZA was an early chatbot.
Expert Systems use a knowledge base and inference engine.
Explainable AI makes AI decisions more understandable.
AI bias may result from biased training data.
Deepfakes are AI-generated or manipulated synthetic media.
Edge AI processes data locally on devices.
AIoT combines AI and IoT.
AGI refers to human-level general intelligence in machines.
ASI refers to intelligence exceeding human capabilities.
Narrow AI is designed for specific tasks.
Overfitting occurs when a model performs well on training data but poorly on new data.
Underfitting occurs when a model is too simple to learn patterns properly.
Transformer models use attention mechanisms.
Tokenisation divides text into smaller units.
Hallucination is the generation of false or inaccurate information by an AI model.
Fine-tuning adapts a pre-trained model for a specific task.
Inference is the process of using a trained model to generate outputs.
Prompt engineering involves designing effective instructions for AI models.

Practice Questions

What is a ‘Neural Network’ in context of Artificial Intelligence?
(A) A type of computer hardware
(B) A software programming language
(C) A model inspired by human brain
(D) A database management system

SDO 2026

Answer: (C) A model inspired by human brain

2. Which of the following is not the domain of Artificial Intelligence?
(A) Creative intelligence
(B) Robotics
(C) Natural Language Processing
(D) Deep learning

SDO 2026

Answer: (A) Creative intelligence

3. Which programming language is most commonly used for Artificial Intelligence and machine learning applications?
(A) Java
(B) Ruby / Lisp
(C) C++
(D) Python

INSP LEG-M (APR 2025) / INV DH(Jan-2024)

Answer: Python

4. The term Artificial Intelligence was coined by
(A) Alan Turing
(B) Paul McCarthy
(C) John McCarthy
(D) Edward Feigenbaum

Liaison Officer (LO) 2025

Answer: (C) John McCarthy
Explanation: Computer scientist John McCarthy is credited with coining the term “Artificial Intelligence” in his 1955 proposal for the seminal Dartmouth Conference.

Pages: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17