Artificial Intelligence full report

[seminar project explorer]

Artificial Intelligence
S.Gokul, F.Ivin Prasanna
3rd Year B.TECH Department of Information & Technology
Adhiyamaan college of Engineering,
Hosur-635109

[attachment=10139]

Abstract
Artificial Intelligence is study and design of machines and design of intelligents, where intelligent agent is a system
tool which prevents its environment and takes actions which maximize its chances of success.
This paper throws light on implementing the artificial intelligence to the word processing and the word editing
software. This can be accomplished by teaching the machine, both the logic & sense that we possess over the
language. It is a very tedious job. Yet it can be achieved through the concept of "Patternisation". It means that the
language is stored as set of rules and regulations, so that the machine can automatically generate its own sentences
and also can easily identify the incorrect ones. This is a key concept which is going to make the ultimate dream of
humans (AI) come true. These ideas are purely unique. This is a research work.

Introduction
It is artificially imparting the intelligence of humans
to the machines, so that they can learn and act by
themselves like us. Major AI textbooks define the
field as "the study and design of intelligent agents",
where an intelligent agent is a system that perceives
its environment and takes actions which maximize its
chances of success.
AI has become the major plot for the fiction writers
since the start of the research. Particularly in the field
of ROBOTICS, AI finds its applications to be
extraordinary. Generally the fiction works are more
exposed than the reality about the AI. In contrast to
fiction, AI has been stuck up in a basic level problem
which is strong enough to get its hold until it is
solved properly. Apart from this hold, it has its
development jailed by many other problems of the
previous breed which is not going to let many
generations enjoy the AI.
In short, we can simply say that it is almost
impossible to replicate the actions of HUMAN
BRAIN artificially using the range of machines
present now. The central problems of AI include
mostly Brain involving activities such as Reasoning,
Knowledge, Planning, Learning, Communication,
Perception and the ability to move and manipulate
objects. General intelligence is the most typical part
to impart.

Problems of AI
We have almost picked all the basic level problems to
develop AI. The basic level problems are emphasized
because there is never a strong building without a
good basement. Let’s discuss all the materials
required for a good basement of AI.
Basic Problems
Lack Of Sense:
AI uses computers because they are the best available
tool, not because they are the object of study. Now
let’s discuss the problems keeping the MS-WORD’s
“Autocorrect” option as the plot. This highly famous
option has never reached the people as a basic and
efficient tool of AI.Coming back to the problems, the
chief problems are concerned with the language. The
stop is in making the computer to understand the
language. We can program the entire language to the
machine but can not make them understand the
language. Let us consider the following situation: We
have programmed the entire English grammar in a
computer with entire grammatical rules and
regulations. Now let us see how the computer reacts
to the following sentences.
1) Cat ate rat
2) Rat ate cat
If we instruct the computer to check the correctness
of the two sentences, the computer will surely give
out that the two sentences are correct. This is a big
drawback. We humans can easily sort out that the
second sentence is grammatically correct but not
sensibly. Because of the sense that we posses we can
understand that the sentence is wrong, but it is merely
impossible to program that sense to the machines.
This is where computers are one step behind the
humans.

For the present time being we have not yet solved
this problem even for the phrases.
Example
1) Leave bus
2) Leave letter
So far, Scientists could not program the machine
even to find the difference in the grammatical sense
of the above sentences. It is very crucial for a
machine to understand the sense in the language.
Now let’s consider the third example.
1) It is midnight.
2) It is dark.
These 2 sentences are correct. But the problem
peeping out is that the computer cannot understand
the coherence between these 2 sentences. Machines
totally ignore that these 2 sentences have some
connection between them.
This leads us to a point that machines do whatever
they are programmed to. They don’t understand and
do. They simply convert the user data into processed
output.
Intelligence
Intelligence is the linking of present events with the
experience and coming out with new ideas. Humans
naturally possess this ability. Our task as discussed
above is imparting this ability artificially to the
machines. I am mentioning this again because there is
a point to note down. The “Linking process” is the
key. Let us consider this example:
I saw TOM& ________.
Ans: ?
Your mind would have answered in a fraction of
second that the answer is JERRY without any
hesitation. Ya. This is intelligence. It becomes
possible for us to do this because of this simple
mechanism. As soon as brain receives a question, it
searches the records saved in it. There is memory of
size more than 1TB in the brain. In a fraction of
second our brain finishes analyzing all these huge
data and finds the answer by linking the given
question to each possibly related data. This is the
mechanism.
Having analyzed the brain mechanism, let us discuss
whether it is possible to achieve this mechanism in
the machines.
Consider our two eyes as a video camera and Brain is
the processing and storage unit. Let us assume that
the eyes capture video for 12hrs a day. Let us assume
that the videos are stored in the standard high quality
AVI format. Let us do a simple calculation.
12hr movie file in std. AVI format= 10GB (approx.)
Movie files stored in 1yr=3650GB
“Can a machine be much efficient to sort the data,
delete the unwanted, remember the most wanted,
process the bulk of data in a fraction of second and
always give us the correct information?”
“What will be the size of storage unit and the
processing speed of the machine if designed so?”
Now you can estimate the might of the brain which
does all these and also more than these and keeps
silent inside us with a negligible weight of 1.5kg.
“Can anyone replicate it artificially at least making
the machine to perform 1/100th of its work?”
For everything, the answer is NO at present.
Betty Crow’s Hook:
Two crows named BETTY and ABEL learnt to use
bent wire to fish a bucket of food from a vertical tube
(as in the picture). Then ABEL flew off with a hook.
• BETTY tried to use a single piece of wire
for a while and then failed.
• The next thing what she did was a great
example for intelligence.
• She then pushed one end of the wire into the
tape holding the tube and moved the other
end using her beak, making a hook.
• She then used the hook to carry the bucket.
• She did this correctly 9 times out of 10.
http:\news.bbc.co.uk\hi\sci\tech\2178920.stm
To find more, give in GOOGLE: Betty crow Hook
This was reported and shown in BBC- August 2002.
This is one of the examples portraying the
intelligence of the living organisms.
This is a simple question to be raised.
“Can a Robot be able to replicate BETTY’s mental
process?”

The answer is No. Machines just do what they are
programmed to. You can even program a machine to
do this work. But it won’t come under AI. AI is more
about making the machines to learn by themselves.
Let us consider the same event and analyze it a bit
differently.
Now let us consider the crow to be a machine. In its
first attempt with a straight wire, it could not produce
the desired output. So it either produces an error
statement or gives out improper output. There is no
chance for it to take efforts to prepare a hook. What
actually happens inside the crow is that it learns that
the output is improper. So it identifies the required
output and changes its code itself to get the required
output.
“Can a machine edit its own code according to the
output?”
The reasons for the behavior of the crow are
1) Innate behavior.
2) Learnt adaptation.
3) Self knowledge.
Now let us see a simple way to overcome this
problem to a certain level.
Patternisation
Mentioning it again and again, A.I is simply making
the machines to replicate the human brain. So let us
discuss how language is covered by brain. It uses the
concept of patternsiation. For example, let us
consider a sentence.
1) Ram is a teacher.
The sentence pattern of this sentence is S+V+C. To
be straight, the brain recognizes the sentences in a
pattern like this. It finds out the error if any part of a
sentence mismatches the pattern. So our task is to
code the pattern and rules to the computer.
Rules cover an important portion in this due to the
flexibility of language. Flexibility leads to a lot of
exceptions which all can be translated to machines as
rules. Whenever there is a special case or an
exception, there must be a rule inserted to maintain
stability.
For example, let us consider the pattern S+V. In this
pattern, a sentence is generated when a subject and a
verb is fed. But the sentence generated is correct only
when the subject is animate. So we need to insert a
rule there which states that
“If subject is inanimate, sentence is wrong.
Else correct”
Machine can thereby generate and check all the
sentences of this pattern. This can be implemented to
all sentence patterns. This is a successful first step.
This is only for sentences level.

Get the report here:
[attachment=10139]