ai_chap_1

What is Artificial Intelligence?

“is the study of how to make computers do things which, at the moment, people do better”

3 types of task domain1. Mundane tasks2. Formal tasks3. Expert tasks

What are some of the tasks that fall under the domain of Artificial Intelligence ?

• 1. Mundane tasks» all those things which (nearly) all of us can routinely do (to various

abilities) in order to act and interact in the world.

» These tasks require complex reasoning

• Perception– vision– speech

• Natural language– understanding– generation– translation

• Commonsense reasoning• Robot control• planning

• 2. Formal tasks

• Games– Chess

– Checkers

• Mathematics– Geometry

– Logic

– Integral calculus

– Proving properties of programs

• 3. Expert tasks» Things that only some people are good at, and which require extensive training.

It can be especially useful to automate these tasks, as there may be a shortage of human experts.

• Engineering

– Design

– Fault finding

– Manufacturing planning

• Scientific analysis

• Medical diagnosis

• Financial analysis

• Expert Systems are concerned with the automation of these sorts of tasks

4 questions

1) What are our underlying assumptions about intelligence?

2) What kinds of techniques will be useful for solving AI problems?

3) At what level of detail, if at all, are we trying to model human intelligence?

4) How will we know when we have succeeded in building an intelligent program?

• The Physical Symbol System: consists of...– symbols which are physical patterns– Symbol structure – a number of instances/tokens of

symbols related in some physical way– Processes which operate on expressions to produce

other expressions.» (creation, modification, reproduction, destruction)

• A Physical symbol system is a machine that produces through time an evolving collection of symbol structures.

1) What are our underlying assumptions about intelligence?

• The PSS Hypothesis:-A PSS has the necessary and sufficient means for general intelligent action.

• Importance of PSS?– Is a significant theory of nature of human intelligence– Forms the basis of belief that if is possible to build

programs that can perform ‘intelligent tasks’

Intelligence requires knowledge, which possesses the following properties…

• it is voluminous.• it is hard to characterize accurately.• It is constantly changing.• It differs from data by being organized in a way that corresponds to the ways it will be used.

What disadvantages does knowledge possess ?

2) What kinds of techniques will be useful for solving AI problems?

There are several Techniques, for the solution of a variety of problems

•The knowledge captures generalizations.Otherwise called as ‘data’

• It can be understood by people who must provide it.• It can easily be modified. -to correct errors –to reflect changes

• it can be used in great many situations.• It can be used to help overcome its own sheer bulk by helping to narrow the range of possibilities that must usually be considered.

AI technique? is a method that exploits ‘knowledge’ that should

be represented in such a way that…

Constraints imposed by AI problems

Introductory Problem: Tic-Tac-Toe

Question answering

Series of solutions Increase in

Their complexity

Their use of generalizations

The clarity of their knowledge

The extensibility of their approach.

1 2 3 4 5 6 7 8 9


X X

o


Blank / X / O

0 / 1 / 2

Program 1:Program 1:

1. View the vector as a ternary number. Convert it to a decimal number.

2. Use the computed number as an index into

Move-Table and access the vector stored there.

3. Set the new board to that vector.


Comments: 1. Efficient in terms of time2. A lot of space to store the Move-Table.3. A lot of work to specify all the entries in the

Move-Table.4. Difficult to extend.



Turn = 1 Go(1)

Turn = 2 If Board[5] is blank, Go(5), else Go(1)

Turn = 3 If Board[9] is blank, Go(9), else Go(3)

Turn = 4 If Posswin(X) 0, then Go(Posswin(X))

.......


Comments:1. Efficient in terms of space2. Not efficient in time, as it has to check several

conditions before making each move.3. Easier to understand the program’s strategy.4. Hard to generalize.(3D tic-tac-toe)


8 3 4 1 5 9 6 7 215 (8 + 5)


Magic squareProgram 2:Program 2:

Comments:

1. Checking for a possible win is quicker.

2. Human finds the row-scan approach easier, while computer finds the number-counting approach more efficient.



1. If it is a win, give it the highest rating.

2. Otherwise, consider all the moves the opponent could make next. Assume the opponent will make the move that is worst for us. Assign the rating of that move to the current node.

3. The best node is then the one with the highest rating.


Comments:

1. Require much more time to consider all possible

moves.

2. Could be extended to handle more complicated

games.


Introductory Problem:Question Answering

“Mary went shopping for a new coat. She found a red one she really liked. When she got it home, she discovered that it went perfectly with her favourite dress”.

Q1: What did Mary go shopping for?

Q2: What did Mary find that she liked?

Q3: Did Mary buy anything?


1. Match predefined templates to questions to generate text patterns.

2. Match text patterns to input texts to get answers.

“What did X Y” “What did Mary go shopping for?”

“Mary go shopping for Z”

Z = a new coat



Structured representation of sentences:

Event2: Thing1:

instance: Finding instance: Coat

tense: Past colour: Red

agent: Mary

object: Thing 1



Background world knowledge:

C finds M

C leaves L C buys M

C leaves L

C takes M


What are 3 important AI techniques ? • Basic techniques that used throughout AI

1) Search

It is the best way so far as no better way has been found to solve the problems and it is also a frame work that direct techniques can be embedded.

2) Use Of Knowledge

It is a way of finding solutions for complicated problems by manipulating the structures of the objects that are involved.

3) Abstraction

Finds a way which separates the more important aspects and modifications from the unimportant ones that would otherwise confuse any process.

What are the advantages of programs that achieve the 3 techniques (Search , Use Of Knowledge and abstraction) over those that do not ?

1) By exploiting these techniques the programs are less delicate. They are less likely to be thrown off completely by a small disturbance in their input.

2) The knowledge of the programs can easily be understood by people.

3) These techniques can be used for larger problems where more direct methods fail.

3) At what level of detail are we trying to model human intelligence?

AI projects are motivated by 2 goals

Goal 1:Trying to produce programs that do the tasks the same way

people do.

Goal 2:Attempting to produce programs that simply do the tasks in whatever way appears easiest.

Building programs that perform tasks the way people do can be divided into 2 classes. What are those 2 classes ?

The first class

Programs, to solve problems that do not really fit our definition of an AI task.

Example: of this class of program is the Elementry Perceiver and Memorizer (EPAM) ,

which its task is to memorize connected pairs of nonsense syllables. Memorizing pairs of nonsense syllables is easy for a computer to do, by just inputing them and to recalling a response syllable by giving its associated syllable and the computer simply scans the associated syllable and responds with the one stored next to it.

In contrast this is a very difficult task for people to do. Many people find this class very uninteresting, however these programs are useful for psychologists who want to test theories of human performance.

The second class

Programs, to solve problems that really fall within our definition of AI tasks

1) To test psychological theories of human performance.

2) To enable computers to understand human reasoning.

3) To enable people to understand computer reasoning.

4) To accomplish what knowledge we can gather from people.

What are the reasons that we might want to model human performance at AI tasks ?

4) How will we know when we have succeeded in building an intelligent program?

``Turing test'' ``Turing test''

The test involves a human communicating with a human and with a computer in other rooms, using a computer for the communication.

The first human can ask the other human/computer any questions they like, including very subjective questions like ``What do you think of this Poem''. If the computer answers so well that the first human can't tell which of the two others is human, then we say that the computer is intelligent.

Human Interrogator

Human

AI System

Imitation Game

ans: YES.

Often it is possible to get a fairly precise and correct measure of the acheivement of a program.

Can the achievement of AI tasks be measured ?

For example: just as human players can possess a CHESS rating, programs can do the same. The chess rating for programs is based on the rating of players whom the program can beat. Already programs have possessed chess ratings higher than the great majority of human players.

For other problem areas, not such a precise measure of a programs acheivement can be measured.

For example: A program called DENDRAL that analyzes organic compounds to identify their structure. It is very difficult to posses exact measure of DENDRAL's level of acheivement in comparison to human chemists, however this program has produced analyses that have been published as original research results hence it is definately performing adequately.

What are AI's most serious dilemmas ?

1) If an AI system wants to handle anything except small problems then it must contain a lot of knowledge.

2) As time goes by, the amount of knowledge keeps growing and it becomes difficult to access the appropriate things when needed so everytime more knowledge must be added to help and then we shall have even more knowledge to handle so we keep adding on more and more knowledge.

AI Programming languages?

LISP (List Processing)

PROLOG (Programming in Logic)

The End

The End

What is AI?

Thinking humanly Thinking rationally

Acting humanly Acting rationally

Acting Humanly: The Turing Test

• Alan Turing (1912-1954)

• “Computing Machinery and Intelligence”

(1950)

Human Interrogator

Human

AI System

Imitation Game

Thinking Humanly: Cognitive Modelling

• Requires testable theories of the workings of the human mind: cognitive science.

(psychology + linguistics + AI)

• Not content to have a program correctly solving a problem.

More concerned with comparing its reasoning steps to traces of human solving the same problem.

Thinking Rationally: Laws of Thought

• Aristotle was one of the first to attempt to codify “right thinking”, i.e., irrefutable reasoning processes.

• Formal logic provides a precise notation and rules for representing and reasoning with all kinds of things in the world.

• Obstacles: Informal knowledge representation.

Computational complexity and resources.

Acting Rationally

• Acting so as to achieve one’s goals, given one’s beliefs.

• Does not necessarily involve thinking.

• Advantages: More general than the “laws of thought” approach.

More amenable to scientific development than human-based approaches.

ai_chap_1

Documents

Transcript of ai_chap_1