Computer CT2 AI

Chapter- "Artificial

Intelligence"

Introduction:

Artificial Intelligence (AI) is gaining the spotlight across applications in our personal

computing lives. The question 'Can a machine think and behave like humans do?' led

to the thought of artificial intelligence. The development of AI started with the intention

of creating similar intelligence in machines that we find and regard high in humans.

Artificial intelligence is composed of two words artificial and intelligence, where

artificial defines 'man-made, and intelligence defines 'thinking power'. Hence AI

means 'a man-made thinking power'.

Definition of Artificial Intelligence:

According to the father of Artificial Intelligence, John

McCarthy, it is 'The science and engineering of making

intelligent machines, especially intelligent computer

programs'.

AI is a form of intelligence; a type of technology and a

field of study. Al theory and development of computer

systems are able to perform tasks that

normally require human intelligence.

It covers a broad range of domains and

applications and is expected to impact

every field.

AI can also be used to extend machine

capabilities to accomplish several tasks

like perceiving, learning, thinking,

decision making and problem solving

referred to as cognitive tasks.

FACT FILE

The term Cognition refers

to 'the use of conscious mental

processes' which essentially

means 'the process of thought.' While

cognitive tasks refers to the actions

which help define cognitive thought

process.

Intelligence:

Intelligence is the ability of a system to reason, learn from experience, solve complex

problems, comprehend new ideas, use natural language fluently, adapt to new

situations, store and retrieve information.

Intelligence is composed of:

* Reasoning

* Problem Solving

* Linguistic Intelligence

* Learning

* Perception

To build an intelligent system in which at least one or all the components of

intelligence can be used. Some examples of intelligent systems are Siri on iPhone and

Alexa by Amazon which uses combinations of the components of intelligence to give

relevant answers.

Goals of Al:

Major goal of an intelligent system is to enable a system to think and behave like

humans do, in order to solve a problem. Intention of AI is to:

1. implement human intelligence in machines by creating systems that can think, act,

learn and behave like humans.

2. create expert systems that can behave intelligently, learn, explain, demonstrate and

give suggestions to its users.

3. enable a system to understand and process natural language.

4. empower a system to perform intellectual tasks that a human can perform.

Application Areas of Al:

AI has been dominant in various fields such as -

* Intelligent Robots: Robots are capable of performing all the tasks that a human

can do. In addition, they are programmed to learn from their mistakes and adapt to

changes.

* Expert Systems: They provide applications which include machines and

programs that provide explanation and advice to humans.

* Gaming: AI plays an important role in the gaming field where machines can think

of large number of possible options based on the data acquired, such as chess,

tic-tac-toe.

* Vision Systems: Using AI, the systems can understand and comprehend the

visual input on the computer.

* Natural Language Processing: AI has gained competency in handling

natural languages spoken by humans.

* Handwriting Recognition: An AI enabled software can read handwritten text

on paper or screen. It can also convert it to editable text.

* Speech Recognition: AI is also capable of handling speech related data. It can

comprehend the commands in different accents. It can also handle background noise

and voice modulation very smartly.

Domains of Al:

The domains of AI are classified as following:

1. Mundane Tasks: These are tasks that we humans do on a routine basis

without any special training. Common sense, planning and reasoning are common

characteristics of this task.

2. Formal Tasks: These are tasks that require formal training like verifications and

theorems. For example, games like chess requires logic building based on theorems.

3. Expert Tasks: They require educational qualification like engineering to make

expert systems that comes under functional expert systems needed at manufacturing

planning, medical diagnosis, etc.

Tools Used for Al:

The following tools are used to imbibe artificial intelligence in computers.

Logic:

Different forms of logic are used in AI research. Logic is used for concept

representation and problem solving. Logic programs are used by the programmers to

verify their correctness. Some of the examples are Proportional Logic which involves

truth functions such as 'or' and 'not'. First Order Logic can express facts about their

objects, their properties and their relations with each other.

Search and Optimisation:

Problems in AI can be solved when we search through the best possible solutions.

Simple exhaustive search methods are used for solving AI problems. To find solutions,

many search techniques use rules of thumb to prioritise choices in favour of those

that are more likely to reach the goal in minimum number of steps, thus optimising the

search techniques.

Classifiers and Statistical Learning Methods:

Classifiers are patterns or observations, where each observation belongs to a class (i.e.

decision to be made). All the classes together forms a dataset. When a new observation

is received, it is based on previous observations in the dataset. Classifiers can be

organised or trained in many ways using various statistical learning approaches.

Probabilistic Methods for Unrealistic Reasoning:

Problems in AI in many stages required to work with assumed data and circumstances.

Al researchers have devised a number of powerful tools to solve these problems using

probability theory and economics. Precise mathematical tools have been developed that

analyse how an agent can make choices and plan using various theories and techniques.

Artificial Neural Network:

Neural as the name suggests is inspired by the human nervous system.

.

It resembles the way:

- the human brain uses 86 billion nerve cells called neurons, which are connected

through thousands of axons.

- the information is received in the form of stimuli through the sensory organs and

accepted by dendrites. These stimuli creates electric impulses, which travel through

the neural system

In the neural network, a neuron can forward the message to other neuron to take care of

the issue.

Artificial neural methods following the same concept receives the input of data, hidden

layer processes the information in a way which the output layer can use. The nodes

(neurals) are connected through links and each link has weight (like the weighted

average). The data is processed using these weights and the output is passed on to

other nodes. The output can be altered by changing the weight value. The artificial

neural method helps in finding patterns in the input which is complex for a human

programmer to identify and teaches the machines to understand.

Problem Solving by Al:

We are surrounded by problems, big or small. Many a time we become so used to a

problem that it becomes a part of our life. Identifying such a problem and having a

vision to solve it, is what problem solving is about (Fig. 9.2).

Problem solving in games like Sudoku can be done by building an artificial intelligent

system to solve the problem. To do so, one needs to define the problem first and then

generate the solution keeping in mind the conditions.

The process of problem solving consists of five steps:

1. Defining the Problem: The problem must be defined explicitly and precisely. It

should contain a number of possible situations to achieve an acceptable solution.

2. Analysing the Problem: Problem analysis and its requirements must be done

to have an impact on the resulting solution.

3. Identification of Solutions: This step generates reasonable amount of solution

to the given problem falling in a particular range.

4. Choosing a Solution: The best solution is chosen from the identified solutions

to achieve the target.

5. Implementing: This is the last step in problem solving after choosing the best

solution. After all the steps you need to implement a solution to solve a problem.

State Space Search:

It is the process of considering states or successive configurations of an instance to find

the goal state from the current state. The given problem is a set of states, where a state

contains all of the information necessary to determine if it is a goal state. The effect of

action determines states and moves from the start state to the goal state.

For example, if a delivery drone is to go from the warehouse address to the delivery

address, it would use the state space search model to do so.

Games are an integral part of our culture. People across the world participate in

different kinds of games as a form of social interaction, competition and enjoyment.

The basic principle of every game is rule setting and following the rules.

Some games that use the AI concept are discussed below-

Rock-Paper-Scissors: A game based on data for AI where the machine tries to

predict the next move of the participant. It is a replica of basic rock, paper and scissors

game where the machine tries to win ahead by learning from the participant's previous

moves.

Mystery Animal: A game based on the natural language processing where the

participant has to guess the animal name by asking maximum 20 questions to an AI

system. The animal randomly gets selected for each game by using an AI system and

the machine replies in either yes or no.

Emoji Scavenger Hunt: A game based on the computer vision where the machine

initiates the game by showing an emoji. The participant is expected to show a similar

object in-front of the camera while the machine keeps on guessing what is being shown

on it.

Limitations of Al:

Despite its advantages, every technology has some limitations as well. The following

are the limitations that we keep in our mind while creating Al systems:

1. Developing Al enabled systems is very expensive.

2. Programming or training these systems is a tough job.

3. There is a lack of feelings, emotions and creativity in these systems.

4. Dependency on machines increases.