ARTIFICIAL INTELLIGENCE CT-1

Part B
1.Explore the horizons of AI and explain in detail.
Firstly, AI may be defined as the branch of computer science that is concerned with the
automation of intelligent behaviour.

1. Healthcare – AI is transforming diagnostics, robotic surgeries, personalized medicine,

and patient monitoring, making healthcare more efficient and precise.

2. Transport – AI powers autonomous vehicles, smart traffic management, and

predictive maintenance, improving safety and reducing congestion.

3. Agriculture – AI-driven drones, automated irrigation, and precision farming optimize

crop yields and reduce resource wastage.

4. Education – AI-powered tutors, personalized learning platforms, and automated

grading enhance student learning experiences.

5. E-commerce – AI improves recommendations, chatbots, and fraud detection,

enhancing customer experience and operational efficiency.

6. Entertainment – AI creates personalized content recommendations, deepfake

technology, and AI-generated media for immersive experiences.

7. Robotics – AI-powered robots assist in industries like manufacturing, space

exploration, and household chores, increasing automation.

8. Social Media – AI moderates content, detects fake news, and personalizes user
experiences through recommendation algorithms.

9. Finance – AI detects fraud, automates trading, and provides predictive analytics for
better financial decision-making.
 10. Gaming – AI enhances realistic graphics, NPC behavior, and personalized gaming
experiences through adaptive AI systems.

2.What is meant by cognitive model?

A cognitive model in AI refers to a computational representation of human cognitive
processes, such as perception, reasoning, learning, problem-solving, and decision-making.
These models aim to simulate how the human mind functions, allowing AI systems to
process information, adapt, and interact in an intelligent manner.

A cognitive model is a framework used to design AI systems that can think and learn similarly
to humans. It is inspired by cognitive psychology and neuroscience, enabling AI to replicate
mental processes such as:

• Perception (understanding visual/auditory input)

• Memory (storing and retrieving information)

• Attention (focusing on relevant data)

• Decision-Making (evaluating options and making choices)

• Language Processing (understanding and generating human language)

Applications of Cognitive Models in AI:

a) Natural Language Processing

Cognitive models enable chatbots, voice assistants (Siri, Alexa), and language translation
tools to understand and generate human-like responses.

b) Computer Vision

AI systems use cognitive models to recognize images, detect objects, and analyze visual data
(e.g., self-driving cars, facial recognition).

c) Robotics & Automation

AI-powered robots use cognitive models to navigate environments, learn tasks, and interact
with humans.

d) Decision Support Systems

AI in finance, healthcare, and business uses cognitive models for intelligent decision-making.

e) Personalized Learning & AI Tutors

AI-powered education platforms use cognitive models to provide adaptive learning

experiences.

Future Applications:
Neuro-Symbolic AI: Combining neural networks and logic-based reasoning for better AI
interpretability.

AI-Driven Consciousness Research: Exploring AI systems that can self-reflect and improve
their own learning.

Brain-Computer Interfaces: Using AI to connect directly with the human brain for enhanced
cognitive abilities.

3.Give the complete solution for 4 queens problem

4.Discuss the important AI techniques

Example: Travelling Salesman and Search Space Problem.

5.What is fully observable or partially observable environment? Explain

Complete Information: The agent can see all relevant aspects of the environment.
No Hidden Variables: There are no unknown factors affecting decision-making.
Perfect State Knowledge: The AI system always knows where it is and what actions to
take.
 Limited Information: The agent can only observe part of the environment at a time.
Uncertainty: Some aspects of the world are hidden, requiring the AI to infer missing
details.
Need for Memory or Probabilistic Reasoning: AI may use history, probability, or models
to predict unknown states.

Crucial Effects of Observability:

1. Decision-Making Strategies – AI in a partially observable environment needs

advanced reasoning, such as reinforcement learning.

2. Complexity of AI Models – Partially observable environments require probability

models and long-term memory, making AI more complex.

3. Real-World Applications – Most real-world environments (healthcare, finance,

autonomous systems) are partially observable, requiring AI to adapt and learn over
time.

6.Illustrate the problem-solving techniques and formulate a problem with an

example.
7.Describe various AI models.

Symbolic AI is a sub-field of artificial intelligence that focuses on the high-level symbolic

(human-readable) representation of problems, logic, and search.

“What is an apple?”, the answer will be that an apple is “a fruit,” “has red, yellow, or green
color,” or “has a roundish shape.” These descriptions are symbolic because we utilize
symbols (color, shape, kind) to describe an apple.
8.Explain in detail the characteristics to be analysed for solving problems in AI
The seven characteristics to analyse problem-solving in AI are:

1.Is the problem decomposable?

→ Some problems can be broken down into smaller sub-problems, solved separately,
and combined to form the final solution.
→ If a problem is decomposable, AI can use divide-and-conquer techniques.
→ If a problem is not decomposable, AI must solve it as a whole.

2.Can Solution steps be ignored or undone?

→ Ignorable: Some steps do not matter in reaching the solution.

→ Undoable (Reversible): The AI can backtrack and undo previous steps.

→ Non-Undoable: Once an action is taken, it cannot be reversed.

3.Is the Universe Predictable?

→ Predictable: The outcome of an action is known (deterministic).

→ Unpredictable: The outcome is uncertain due to randomness (stochastic).

4.Is a good solution absolute or relative?

→ Absolute Solution: There is a single correct answer.

→ Relative Solution: The best solution depends on the situation and constraints.

5.Is the solution a state or a path?

→ State-based problems: The final state matters, not how AI got there.
 → Path-based problems: The steps taken to reach the goal are important.

6.What is the role of knowledge?

Some problems require prior knowledge, while others can be solved purely by reasoning
or search.

7.Does the task require interaction with a person?

Some AI tasks require human interaction, while others operate independently.

Example with Chess Problem:

9.Analyse the significance of each phase of Data Acquisition Process and
explain with a real time example.
PART C
1.You are given two jugs, a 5-gallon one and a 3-gallon one, a pump which has
unlimited water which you can use to fill the jug, and the ground on which
water may be poured) Neither jug has any measuring markings on it. How can
you get exactly 4 gallons of water in the 5-gallon jug? Formulate this problem
and solve.
Formulation of the Problem

This is a water jug problem, a classic example of a state-space search problem in AI and the
objective is to measure exactly 4 gallons of water using only a 5-gallon jug and a 3-gallon jug.

Problem Representation

• State Representation: (x, y), where

• x = amount of water in the 5-gallon jug

• y = amount of water in the 3-gallon jug

• Initial State: (0,0) → Both jugs are empty.

• Goal State: (4, y), where x = 4 (any value for y).

• Allowed Actions:
1. Fill a jug completely from the pump.

2. Empty a jug onto the ground.

3. Transfer water from one jug to another until either the receiving jug is full or the
pouring jug is empty.

(0,0)

↓ Fill 5-gallon

(5,0)

↓ Transfer to 3-gallon
(2,3)

↓ Empty 3-gallon

(2,0)

↓ Transfer from 5-gallon to 3-gallon

(0,2)

↓ Fill 5-gallon

(5,2)

↓ Transfer to 3-gallon

(4,3) → Goal Reached!

2.Describe the AI characteristics for the following problem: Teaching a Toy to

Recognize Shapes. Suppose you have a toy that can identify basic shapes
(circle, square, and triangle). This toy needs to be programmed to recognize
and categorize different shapes.
1. Is the problem decomposable?

Yes, the problem can be decomposed into smaller subproblems, such as:

→ Feature extraction: Identifying key features of shapes (e.g., angles, edges,

curves).

→ Classification: Categorizing the shape based on extracted features (e.g., is it a

circle, square, or triangle?).

→ Learning: Using labelled data (shapes with known categories) to teach the toy
to recognize new shapes.
These subproblems are manageable and can be solved individually to form a
complete solution.

2. Can solution steps be ignored or undone?

No, each step in teaching the toy is important. Ignoring or undoing previous steps could lead
to incorrect recognition or misclassification of shapes. For example, skipping the feature
extraction phase may prevent the toy from distinguishing between different shapes.

3. Is the Universe Predictable?

Yes, but with some assumptions. The problem's universe is largely predictable if we assume:

→ The shapes are well-defined (perfect geometrical shapes with no distortion).

 → The toy uses a defined set of features (angles, sides, etc.) to recognize shapes.
However, external factors like sensor noise, incorrect shape presentations, or unusual
shapes (e.g., distorted versions) could introduce unpredictability.

4. Is a good solution absolute or relative?

Relative, the definition of a "good solution" is relative here. The toy doesn't need to achieve
perfect recognition, but rather recognition that is accurate enough within a given threshold
(e.g., 90% accurate). The solution could vary based on:

→ The variety of shapes to be recognized.

→ The tolerance level for classification errors.

→ The complexity of the learning algorithm used.

5. Is the solution a state or a path?

State, the solution is primarily a state—once the toy recognizes a shape (e.g., circle, square,
or triangle), it has reached a final classification state. There isn’t a need for a continuous
"path" of intermediate steps beyond the initial recognition; once a shape is identified, the
task is complete.

6. What is the role of knowledge?

High importance, knowledge plays a crucial role in the shape recognition task:

→ Feature knowledge: Understanding the properties that define each shape (e.g., a
square has four equal sides, a triangle has three sides, and a circle has no corners).

→ Training data: The toy needs a dataset of labelled shapes (with corresponding
features) to learn the patterns and recognize new shapes.

→ Algorithmic knowledge: The toy needs knowledge of how to apply machine learning
algorithms (e.g., decision trees, neural networks, etc.) for classification.

7. Does the task require interaction with a person?

Yes, during the learning phase. Initially, the toy will need to interact with a person to learn
from labelled examples (supervised learning). The person might provide the toy with
different shapes, ensuring it learns to classify them correctly. However, once the toy is
trained, it may not need interaction unless it requires corrections or additional training data.

3. Examine the problem type for the following and explain

a.Vacuum cleaning agent without sensor
b.Vacuum cleaning agent with sensor
c.Google-search
d.Tic-tac-toe
e.Travelling salesman
a. Vacuum Cleaning Agent without Sensor

→ Non-Deterministic or Partially Observable

→ The environment is partially observable because the vacuum cleaning agent cannot
sense the state of the environment (whether a specific space is dirty or clean).

→ The agent's actions are deterministic (e.g., moving left or right, cleaning a spot), but
without sensors, it lacks real-time feedback to adjust its actions.

→ The agent cannot see where it has already cleaned, leading to multiple possible
states. Hence, it operates with incomplete information.

b. Vacuum Cleaning Agent with Sensor

→ Type: Deterministic or Observable (Single-State)

→ The environment is now observable because the agent has sensors to detect dirt or
cleanliness in its surroundings.

→ The state is deterministic because the agent can predict the outcome of its actions (if
it cleans a dirty spot, it becomes clean).

→ The agent can make decisions based on its observations, so there's a single state to
determine the next action (if dirty, clean; if clean, move).

c. Google Search

→ Type: Non-Deterministic or Partially Observable

→ The environment is non-deterministic because the outcomes of the search results

depend on various factors that can change over time, such as the evolving state of
web pages or the user’s location.
→ Partially observable in the sense that Google has access to indexed data but cannot
fully predict the user’s intent based only on a query. The search results are
influenced by algorithms, personalization, and real-time data, making it harder to
know in advance the exact results.
→ The agent (Google search) tries to generate the best results from a vast, evolving
state space, with the exact outcome not fully known until the search is complete.
d. Tic-Tac-Toe

→ Type: Deterministic or Observable (Single-State)

→ The environment is fully observable since both players can see the entire 3x3 grid at
all times.
→ The problem is deterministic because each move made by a player directly changes
the state of the game (placing X or O), and the outcome is predictable.
→ There's only a single state at any point in the game that can be evaluated (who won,
who needs to make the next move), and no hidden or uncertain factors affect the
game.

e. Traveling Salesman Problem (TSP)

→ Type: Deterministic or Observable (Single-State)

→ The environment is fully observable because the cities and distances between them
are known upfront.
→ The problem is deterministic because, given a set of cities and distances, the path
taken by the salesman is predictable (although there are many possible paths).
→ The state of the problem is well-defined: the salesman’s current location and the
cities yet to be visited, with no ambiguity in the environment.
→ Once the cities and distances are provided, there's a single state space to explore.

4.The 8-puzzle, an instance of which is shown below, consists of a 3×3 board

with eight numbered tiles and a blank space. A tile adjacent to the blank
space can slide into the space. The object is to reach a specified goal state,
such as the one shown on the right of the figure. Formulate this problem and
solve.
Formulating the 8-Puzzle Problem

State Representation:

A state is represented as a 3x3 grid. The grid consists of 9 positions (indexed from 0 to 8),
with each position holding a value from 0 to 8.

The value 0 represents the blank space.

The tiles numbered 1-8 represent the numbered tiles in the puzzle.

Initial State:

The initial state is a random configuration of the tiles (except the goal state configuration).

Actions:

The actions consist of sliding one of the tiles adjacent to the blank space (up, down, left, or
right) into the blank space. A tile can only move if it is adjacent to the blank space.

There are 4 possible actions from any position, but some may be restricted based on the
position of the blank space.

Successor Function:

The successor function generates all possible states that can be reached from a given state
by sliding a tile into the blank space. For each configuration of the board, the possible moves
can be computed based on the position of the blank space.

Goal Test:

The goal is reached when the state of the puzzle matches the goal configuration.

Path Cost:

The cost of each move is typically considered uniform (e.g., each move has a cost of 1).
5.Give PEAS description for online bus reservation system
The PEAS (Performance Measure, Environment, Actuators, and Sensors) framework is used
to define the working of an intelligent agent. Here’s how it applies to an Online Bus
Reservation System:

1. Performance Measure:

The system's effectiveness is evaluated based on:

→ Accuracy of seat availability updates

→ Fast and secure booking process

→ User satisfaction and ease of use

→ Minimal booking errors or double bookings

→ Secure payment processing

→ Timely notifications and updates

2. Environment:
The system operates in an environment consisting of:

→ Users (passengers, agents, admins)

→ Online databases of bus schedules, seats, and payments

→ Payment gateways for transactions

→ Web and mobile interfaces for booking

→ External APIs (e.g., maps, traffic data)

3. Actuators:

The system performs actions through:

→ Updating seat availability in real-time

→ Processing bookings and payments

→ Sending notifications (confirmation, cancellation, reminders)

→ Assigning seats dynamically

→ Updating travel schedules based on real-time data

4. Sensors:

The system collects data from:

→ User input (search queries, payment details)

→ Database updates (seat availability, bus schedules)

→ Payment verification systems

→ GPS and traffic data for bus tracking

→ Feedback and reviews from users

6.Formulate the problem for the following and analyse their problem types
a.CHESS game ; b.Blocks world
a. Chess Game

1. State Representation:

→ A state in chess is just the arrangement of pieces on the board at any point in
time.
→ For example, a state would show where each piece (like a pawn, knight, or
queen) is placed on the 8x8 board. It also keeps track of whose turn it is (white or
black).
2. Initial State:

→ The initial state is the starting position in a chess game, where the pieces are in
their standard places at the beginning of the game.

→ White pieces are in the first two rows, black pieces are in the last two rows, and
both players have their pieces in the correct spots to start.

3. Actions:

The actions are the moves players make. For example, a pawn can move one square
forward, or a knight can move in an "L" shape. Each piece has specific rules for how it
can move.

4. Transition Function (What happens after a move?):

The transition function is how the board changes after a move. If a player moves a
piece, the board updates to reflect that move. For example, if a knight moves, the
new board layout will show the knight in its new position.

5. Goal State:

The goal in chess is to checkmate the opponent’s king, which means you put their
king in a position where it can't escape capture. When that happens, the game is
won.

6. Path Cost:

Path cost is how we count the number of moves in the game. In chess, each move
counts as one action, so the cost is simply the number of moves made to reach a goal
state (e.g., checkmate).

Problem Type:

→ Partially Observable: Players can see the current state of the board, but they cannot
predict the opponent’s future moves.

→ Non-deterministic: Chess is non-deterministic because the outcome depends on both

players' unpredictable decisions, and the game can take many different paths.

b. Blocks World

1. State Representation:

A state in Blocks World shows where each block is and whether it's stacked on
another block or sitting on the table. It also tells whether a block is clear (meaning
nothing is on top of it).

2. Initial State:
 The initial state is when all the blocks are on the table, and no blocks are stacked on
top of each other. The blocks are also "clear," meaning you can stack things on top of
them.

3. Actions:

The actions in this world involve moving blocks around:

→ You can stack one block on top of another.

→ You can unstack a block from another block.

→ You can also pick up a block and place it on the table.

4. Transition Function (What happens after an action?):

The transition function describes what happens when you perform an action. For
example, if you stack one block on top of another, the state changes, and the new
configuration shows that one block is now stacked on top of the other.

5. Goal State:

The goal state is a specific arrangement of blocks. For example, the goal might be to
have Block A on top of Block B and Block B on top of Block C, or any other
arrangement specified at the start.

6. Path Cost:

Path cost refers to how many actions you need to take to get to the goal state. Each
move, like stacking or unstacking a block, has a cost of one. So, the total path cost is
the total number of actions needed to reach the goal.

Problem Type:

→ Deterministic: Each action (like stacking or unstacking blocks) leads to a

predictable result.
→ Fully Observable: The agent has complete knowledge of the current state,
including where each block is and whether it’s clear or stacked.