00:00:00
Hello, this is Sheila Dvi. I am the
00:00:03
instructor for your course on artificial
00:00:07
intelligence. So this is the first
00:00:09
lecture in this course and um I will
00:00:14
just do a small introduction before I
00:00:17
start the course. So this will talk
00:00:20
about uh some introductory concepts of
00:00:24
this course. So basically uh the
00:00:27
textbooks which need to be followed are
00:00:30
as follows. The first textbook is
00:00:34
artificial intelligence book by Russell
00:00:37
and Norwick which uh is the one which
00:00:40
has been mostly followed in this course.
00:00:43
Another book is the artificial
00:00:45
intelligence book by Luger which is used
00:00:48
for some small specific topics here and
00:00:51
there and there are some other textbooks
00:00:54
also which have been followed and I will
00:00:56
let you know from time to time what are
00:00:58
the textbooks that I have used over
00:01:00
there. So what are the broad topics
00:01:03
which we will be covering in this
00:01:05
course. So basically we'll start off
00:01:08
with talking about what are agents and
00:01:11
uh uh what are intelligent agents and
00:01:14
the different types of intelligence
00:01:16
agents that we have. So we have
00:01:18
something called a simple reflex agent
00:01:21
and we also have more uh complex agents
00:01:25
okay which are more powerful. So we'll
00:01:29
talk about different types of agents and
00:01:32
an important topic in artificial
00:01:35
intelligence is the general problem
00:01:36
solving. In other words, if you have a
00:01:39
problem, how would you have a general
00:01:41
procedure for solving the problem? So
00:01:44
usually you use like a search algorithm.
00:01:48
So you search for a solution to the
00:01:50
problem. So this is something like the
00:01:52
way human beings do the uh problem
00:01:56
solving. And the next topic will be on
00:01:59
logic. So what are the different types
00:02:01
of logic we have? Uh mainly you have
00:02:04
things like propositional logic u etc.
00:02:08
And when we represent things in logic
00:02:11
then how do we carry out reasoning using
00:02:14
that logic? Because obviously when we
00:02:17
are talking artificial intelligence when
00:02:20
you have certain knowledge you're trying
00:02:22
to find new knowledge by using the means
00:02:25
of reasoning and also you have things
00:02:27
like uh theorem proving which will tell
00:02:30
you whether you know something is right
00:02:33
or wrong etc. And then we also have
00:02:36
other types of uh logic like fuzzy
00:02:38
logic, non-monotonic logic etc. So we
00:02:42
will talk about all those types of
00:02:44
logic. Another uh topic is planning. So
00:02:49
how do you find a plan to carry out a
00:02:52
certain goal? For example, if you have
00:02:55
some space mission, so what are the
00:02:58
steps which need to be taken in the
00:03:00
space mission? So you plan it out. So
00:03:04
that is what we mean by planning. So
00:03:07
there are different types of planning
00:03:09
algorithms. So we'll go through those
00:03:11
different planning algorithms and as we
00:03:15
know learning is a very important
00:03:17
concept over here. So we have uh things
00:03:20
like um supervised learning uh
00:03:23
unsupervised learning, reinforcement
00:03:26
learning and in each of these for
00:03:28
example supervised learning you have
00:03:30
things like um different machine
00:03:32
learning algorithms, deep learning
00:03:34
algorithms. So we'll be going through
00:03:36
all those type of things and uh when we
00:03:39
come to um languages of course along
00:03:44
with the usual languages which are there
00:03:47
we also have some AI languages which are
00:03:50
used. So these AI languages are are
00:03:52
things like uh list, prologue etc. where
00:03:56
u you know some concepts for example
00:04:00
list processing and processing of text
00:04:04
etc is carried out and and prologue
00:04:07
which you could call as something like a
00:04:09
shell expert system shell. So we'll be
00:04:12
going through some of these languages
00:04:14
and communication of course the agent
00:04:16
needs to communicate with the human
00:04:18
being or communicate with each other. So
00:04:21
how this communication is carried out.
00:04:23
So that is what we mean by natural
00:04:26
language processing and we'll talk about
00:04:28
other things like multiple agents, web
00:04:30
agents, negotiating agents and all these
00:04:34
uh different types of
00:04:36
agents. Okay. So Okay. So basically what
00:04:41
are we doing in artificial intelligence?
00:04:43
We are trying to model how the human
00:04:46
being carries out a task. In other
00:04:49
words, you're trying to model human
00:04:52
cognition or the mental faculty which
00:04:55
the human being carries on and then
00:04:58
comes to a decision and and carries out
00:05:00
that decision. So you're trying to make
00:05:03
the computer do things which the human
00:05:05
being does now and it is considered to
00:05:08
be something intelligent. Okay. So you
00:05:12
want the computer to to do things which
00:05:15
require intelligence basically. So
00:05:17
that's the basic idea over here. Okay.
00:05:21
So let us go into some definition of
00:05:24
artificial
00:05:25
intelligence. So we can say that it's a
00:05:28
branch of computer science. Okay. So AI
00:05:31
is considered to be a branch of computer
00:05:33
science where you're trying to uh study
00:05:36
and create computer systems that exhibit
00:05:40
some intelligence. Okay. Or which we
00:05:42
associate with intelligence in human
00:05:45
behavior. So when a human be behavior
00:05:48
does um even things like recognizing a
00:05:51
person or even a child recognizing a
00:05:54
person you can say that it is using its
00:05:57
intelligence, right? So that
00:05:59
intelligence we are trying to model by
00:06:02
using a computer basically. Okay. So in
00:06:06
other words you're trying to find a
00:06:08
machine which thinks. Okay. So we want
00:06:13
to classify a machine as thinking or in
00:06:16
other words if it needs intelligence to
00:06:19
do something then we say it's a thinking
00:06:22
machine right? So we are trying to uh
00:06:25
create that an artificial intelligence.
00:06:28
So here are some different definitions
00:06:30
given in different places. So behavior
00:06:33
of a machine if when performed by a
00:06:36
human being would be considered
00:06:39
intelligent. Another is study of how to
00:06:41
make computers do things which at the
00:06:44
moment people are better. So what people
00:06:47
do better now okay how do you make a
00:06:50
computer do the same thing? The another
00:06:54
one is theory of how the human mind
00:06:56
works. Okay. So you try to see how the
00:06:59
human mind works and you try to simulate
00:07:02
that to get the same result. So these
00:07:05
are different definitions of artificial
00:07:09
intelligence basically. Here are some
00:07:11
more definitions from different places.
00:07:15
So the famous Alan Turing. So you have
00:07:18
the Turing test which we will talk about
00:07:20
little later. So actions that are
00:07:23
indistinguishable from a humans. Okay.
00:07:26
So if the machine carries out a task
00:07:29
which looks like it's been carried out
00:07:31
by a human being. So AI is the study of
00:07:35
complex information processing problems
00:07:38
that often have their roots in some
00:07:40
aspect of biological information
00:07:43
processing. The goal is to identify
00:07:45
solvable and interesting information
00:07:48
processing problems and solve them. So
00:07:50
this is David Maher from MIT.
00:07:54
uh the intelligent connection of
00:07:56
perception to action. In other words,
00:07:59
the computer perceives something and
00:08:02
then after perceiving something, it
00:08:04
needs to take an action. So that action
00:08:07
which it takes, how it comes to decide
00:08:10
what action to take, that is where there
00:08:14
is a intelligence which is involved.
00:08:18
So when we are talking about uh
00:08:21
artificial intelligence, these things
00:08:23
are some of the things which need to be
00:08:26
done. Learning new concepts and tasks.
00:08:29
Okay. Reason and draw useful conclusions
00:08:32
about the word world. Remember
00:08:34
complicated interrelated facts and draw
00:08:37
conclusions from them. So that is
00:08:39
something like the expert systems which
00:08:41
we talk about. understand a natural
00:08:44
language or perceive and comprehend a
00:08:46
visual
00:08:47
scene. Look through the camera and see
00:08:50
what is there. Okay. And move around. So
00:08:53
that is something like what happens in
00:08:55
robotics basically. Okay. Plan sequence
00:08:59
of action. So that is what we call
00:09:01
planning. How do you plan? For example,
00:09:03
you want to go u and buy some milk and
00:09:08
oranges and bananas say right. So what
00:09:10
is the series of actions that you're
00:09:13
going to carry on? So you'll have to
00:09:14
have something like leave the house, go
00:09:17
to the supermarket, buy some milk, go to
00:09:20
the hardware store, buy something etc.
00:09:22
Right? And then come back home. So what
00:09:25
you have is a series of actions to carry
00:09:28
out a task. Okay? Next is offer advice
00:09:32
based on rules and situations. So that
00:09:34
is like the expert system. So you may
00:09:36
have an expert system which will um if
00:09:39
you give the symptoms of a person it
00:09:42
tells you what disease they have. Okay.
00:09:44
So that's what we mean by offering
00:09:47
advice. And of course it may not
00:09:49
necessarily imitate human uh thought
00:09:52
processes in in some cases it could be
00:09:55
done in a different way but get the same
00:09:57
result
00:09:59
basically. Okay. Uh perform tasks that
00:10:03
require high level of intelligence. etc.
00:10:06
So that's what you're trying to see how
00:10:08
to make the computer do basically. So
00:10:11
here is
00:10:13
uh an overall block diagram of
00:10:16
artificial intelligence. So if you see
00:10:18
in the middle it tells you all the
00:10:20
subjects that are covered under
00:10:22
artificial intelligence. So we spoke
00:10:24
about reasoning, learning, planning,
00:10:28
right?
00:10:29
um knowledge uh acquisition and all
00:10:33
those other things right and of course
00:10:36
there's a lot of int u uncertainty
00:10:38
involved in that and then we have some
00:10:41
parent disciplines so the parent
00:10:43
disciplines are the disciplines from
00:10:45
where artificial intelligence has
00:10:47
actually come up okay for example
00:10:49
philosophy psychology maths physics
00:10:53
computer science
00:10:54
etc and if you look down you can see
00:10:58
some application areas of AI. For
00:11:00
example, game playing. So, game playing
00:11:02
is a very important um application
00:11:05
areas. So, things like chess, checkers,
00:11:09
you know, all these types of
00:11:11
games, theorem proving, okay, language
00:11:14
understanding,
00:11:16
uh robotics, etc. So, these are some
00:11:19
application areas which we are talking
00:11:21
about. Okay. So here we can see the
00:11:24
different um topics which will be
00:11:28
covered under AI
00:11:31
basically. Okay. So what are these
00:11:34
topics? You have things like uh
00:11:37
knowledge representation. So how do you
00:11:39
represent knowledge? Okay. So that's a
00:11:41
very important concept. So when we talk
00:11:44
about um say ordinary data structures,
00:11:48
algorithms etc. we talk about data.
00:11:50
Okay. So you have data to solve a
00:11:53
problem. Whereas here you talk about
00:11:56
knowledge. For example, you can have uh
00:11:59
knowledge in your uh database like all
00:12:02
crows are black. Okay. So that would be
00:12:05
like a knowledge. So um you need to do
00:12:10
representation of knowledge. Okay. Not
00:12:12
only uh facts. Okay. So you need to have
00:12:15
facts and rules etc. theorem proving,
00:12:18
game playing, common sense reasoning.
00:12:22
Okay. Uh learning models, inference
00:12:25
techniques, pattern recognition. Okay.
00:12:27
So, some classification algorithms,
00:12:30
searching and matching etc. and logic.
00:12:34
So, you have things like fuzzy logic,
00:12:36
temporal logic, model logic, all these
00:12:39
things.
00:12:41
planning and other sub areas are like
00:12:43
natural language understanding, computer
00:12:46
vision, understanding, spoken
00:12:48
utterances, okay, intelligent uh
00:12:51
tutoring systems etc. Okay, these are
00:12:54
all some sub areas which will be
00:12:56
basically covered. So if we come to
00:12:59
intelligence obviously when we say that
00:13:02
artificial intelligence is when the
00:13:05
computer does something intelligent we
00:13:07
need to be able to define what is
00:13:09
intelligence. Okay. So let us now take
00:13:11
up the definition of intelligence. So
00:13:15
when we come to intelligence if you look
00:13:17
at the dictionary this is what the
00:13:19
dictionary says about intelligence. The
00:13:22
ability to learn and understand or to
00:13:25
deal with new or trying situations.
00:13:27
Okay. So if you have a new situation,
00:13:29
how do you deal with that? Okay, so that
00:13:31
is where you need intelligence and how
00:13:34
do you reason the skilled use of reason.
00:13:38
Another one is the ability to apply
00:13:40
knowledge to manipulate the environment.
00:13:43
Okay? Or to think abstractedly as
00:13:45
measured by objective criteria, the
00:13:48
capacity to learn and solve problems.
00:13:51
The ability to act rationally. So these
00:13:54
are all uh the definition which is given
00:13:57
in the dictionary. Okay. So basically
00:13:59
it's the capacity to learn and solve
00:14:03
problems. Okay. So the ability to solve
00:14:05
novel problems. Okay. If you're given a
00:14:08
new problem for which you don't have any
00:14:11
solution what the ability to do that and
00:14:14
the ability to act in a rational manner.
00:14:16
So rational manner means in a manner
00:14:19
which is gives the right decision
00:14:21
basically. Okay, the ability to act like
00:14:24
the way human beings act. Okay, so this
00:14:27
is what you mean by intelligence and
00:14:29
artificial intelligence is to build and
00:14:32
understand intelligent entities or
00:14:35
agents. Okay, and there are two main
00:14:37
approaches like you say engineering
00:14:39
versus cognitive modeling. So basically
00:14:43
there are things like um you know the
00:14:46
engineering part of it for example the
00:14:49
uh robotics etc. And then we have the
00:14:52
cognitive modeling. Okay. So which we do
00:14:55
by programs etc. So we will not be going
00:14:58
into the robotics part over here.
00:15:01
Basically we are seeing it from the
00:15:03
computer science point of
00:15:06
view. Okay. So what is it that you
00:15:09
require? Okay. In
00:15:11
u the case of uh you know learning or or
00:15:16
or intelligence right? So if you want to
00:15:19
act in an intelligent way, what are the
00:15:22
different uh uh characteristics that
00:15:25
should be there in your AI system? It
00:15:28
should be able to interact with the real
00:15:30
world. So how do you do that? Maybe you
00:15:33
can have some sensors. So if you take
00:15:35
human beings, we have sensors like eyes,
00:15:39
nose, touch, right? Mouth, whatever all
00:15:43
these things. Um so the in the case of
00:15:46
the computer also you need to have
00:15:48
sensors so that it's able to interact
00:15:51
with the real world and find out what is
00:15:54
happening in the real world. So what you
00:15:57
need to do is to perceive to understand
00:15:59
and to act.
00:16:01
Okay. So um so here uh ability to take
00:16:06
action. Okay. So you need to understand
00:16:08
what is happening in the in the real
00:16:10
world and you need to take action for
00:16:13
that reasoning and planning. So you
00:16:16
you're modeling the external world given
00:16:19
some input solving new problems, ability
00:16:22
to deal with unexpected problems etc.
00:16:25
Then learning and adaptation. We need to
00:16:28
continuously learn and adapt. Okay. So
00:16:30
as the real world changes, we need to
00:16:33
also adapt ourselves according to what
00:16:36
is happening in the real world. Um our
00:16:40
models have to be always updated. Okay.
00:16:45
So that is uh these are the things which
00:16:48
are necessary for basic intelligence.
00:16:51
Okay. So these are some people who have
00:16:54
really been pioneers when it came to
00:16:57
artificial intelligence. So you find
00:16:59
many of them were working in the 40s50s
00:17:03
and60s okay 1940s50s
00:17:06
and60s. So the first is Makati. So the
00:17:10
first thing he's known for is lisp.
00:17:12
Okay. So lisp is a list processing
00:17:16
language and um it it was uh started in
00:17:20
around 1960. So he came up with it
00:17:23
around 1960 and uh it it carries out
00:17:26
some type of logical programming and
00:17:29
non-monotonic. He's also known for
00:17:31
non-monotonic reasoning and all these
00:17:33
things. Next one is Minsky. So he was
00:17:37
the founder of AI lab at MIT and uh
00:17:41
there's a knowledge structure called
00:17:43
frames. So he is responsible for that
00:17:46
particular knowledge structure. Then you
00:17:49
have Simon. Uh so Simon is known for the
00:17:53
general problem solver. So that's a very
00:17:55
important
00:17:57
uh uh you know software he came up for
00:18:01
uh solving general
00:18:03
problems. Uh and of course he's worked
00:18:06
on social networks and come up with the
00:18:08
power law which is another important
00:18:10
thing. Then you have Arthur Samuel. So
00:18:13
author Samuel
00:18:15
um term coined the term machine learning
00:18:18
and he was known
00:18:20
for using AI for game playing. So in
00:18:23
other words he was responsible for all
00:18:25
the checker programs which were written
00:18:27
in the
00:18:29
1950s. So he wrote a series of um
00:18:33
checkers programs basically. Then you
00:18:36
have Alan Newell who came up with what
00:18:38
is called the logic theorist which is
00:18:40
logic based uh reasoning general problem
00:18:45
solver. Okay. And and he was responsible
00:18:48
for what is called the chess machine.
00:18:51
Then you have Neielson. So you have the
00:18:53
AAR algorithm. So the AAR algorithm is a
00:18:56
very important huristic search algorithm
00:19:00
which we will be talking about later. So
00:19:02
he is the one who came up with that
00:19:04
algorithm and uh he was he also talk
00:19:07
about planning and um uh he there's a
00:19:10
book by written by him on artificial
00:19:13
intelligence okay which is also a very
00:19:15
well-known
00:19:17
book. So um we'll now talk about what is
00:19:21
called the Turing test. So what is a
00:19:23
Turing test? So Turing test also called
00:19:26
the imitation game. In other words, you
00:19:29
have a human interrogator who is
00:19:32
separated from
00:19:34
uh the AI system, okay, or the computer
00:19:38
system, okay? And there's a teley type
00:19:40
with which he communicates with the AI
00:19:44
system. And if the interlocator cannot
00:19:47
tell whether the whether it is a
00:19:49
computer or it is a human being, okay,
00:19:52
at the other end, the computer actually
00:19:54
passes the test. So that is what is
00:19:57
known as the Turing test. Okay. Now what
00:20:00
are the capabilities that the AI system
00:20:03
needs to have to pass the Turing test?
00:20:05
That's what we have to see. So the one
00:20:08
is that it should be able
00:20:10
to work with natural language
00:20:13
processing. So it should be able to
00:20:15
communicate with the interrogator. So
00:20:17
the interrogator asks some questions. So
00:20:20
it should be able to understand that and
00:20:22
it should be able to reply to those
00:20:25
questions in natural language
00:20:26
processing. Next is the knowledge
00:20:29
representation. So whatever information
00:20:32
it has, it should be able to store it,
00:20:34
okay? So that it has all those
00:20:37
information when it needs to answer
00:20:39
questions, okay, or communicate with the
00:20:43
interrogator. Third one is what is
00:20:45
called the automated reasoning. So when
00:20:48
the interrogator asks a question, it
00:20:51
should be able to reason out using the
00:20:54
information or the knowledge it has
00:20:57
stored. Okay. And it should be able to
00:21:00
draw conclusions, answer questions etc.
00:21:03
And the fourth one is machine learning
00:21:05
where it needs to adapt to new
00:21:08
circumstances. So these are some of the
00:21:11
capabilities required by a computer if
00:21:14
it needs to pass what is called the
00:21:16
Turing
00:21:19
test. Okay. So let us see um uh it's a
00:21:24
multidisciplinary subject AI. Okay. So
00:21:27
number of different areas are involved
00:21:30
here. First one is of course the
00:21:32
computer engineering. Okay. So obviously
00:21:35
the artifact we are using is the
00:21:37
computer right? So the computer is an
00:21:40
important uh aspect over here. So
00:21:43
computer science, data science that is
00:21:46
storing of the information in the data,
00:21:48
machine learning, deep learning, all
00:21:50
these things come under the computer
00:21:53
science and then you have uh subjects
00:21:56
like philosophy, psychology, cognitive
00:21:58
science. Okay. So these are another
00:22:01
important uh uh set of u subjects. Okay.
00:22:07
Another is uh mathematics, physics etc.
00:22:10
So mathematics of course things like
00:22:12
logic what we are using logic reasoning
00:22:16
all these things will come under
00:22:18
mathematics and of course things like
00:22:20
linguistics when you do natural language
00:22:22
processing robotics where um you have
00:22:26
the machine
00:22:27
um you know moving from one place to
00:22:30
another carrying out tasks etc. So that
00:22:33
is robotics. So all these areas will
00:22:38
come under AI. Okay. So all these are
00:22:40
something like the parent disciplines of
00:22:43
AI. Okay. So let us see how these
00:22:47
different topics are used over here.
00:22:49
First of all, philosophy. If you
00:22:52
consider
00:22:53
um logic of course also comes under
00:22:57
philosophy. You had Aristotle working on
00:23:00
some type of logic in in the earlier
00:23:03
parts. Okay. Uh methods of reasoning and
00:23:07
looking at mind as a physical system. Uh
00:23:10
foundations of learning, language,
00:23:12
rationality, language, rationality, all
00:23:15
these things come under
00:23:17
philosophy. And then uh if you look at
00:23:20
mathematics, you have formal
00:23:22
representation and proofs, algorithms.
00:23:25
Okay. So here what happens is you have
00:23:28
algorithms which are intractable
00:23:30
basically okay because you're talking
00:23:32
about really big um data etc. So how do
00:23:37
you work on these intractable problems
00:23:39
then you have things like
00:23:42
decidability computation all these
00:23:44
things then probability and statistics
00:23:47
obviously you have lot of uncertainty
00:23:50
involved. So how do you model this
00:23:52
uncertainty? One way is by using
00:23:55
probability or one of the important ways
00:23:59
is by using
00:24:01
probability. Okay. So these uh that's
00:24:04
another area. Then economics. So in
00:24:07
economics obviously when you make a
00:24:09
decision. Okay. So how do you take a
00:24:13
decision? Okay. So the the to taking a
00:24:15
decision is based on decision theory and
00:24:19
um when the there is more more than one
00:24:22
method of doing it you can use what is
00:24:24
called the utility theory. So what is
00:24:26
the method which gives the highest
00:24:29
utility among the uh decision uh among
00:24:33
the different um actions that you can
00:24:36
take. Okay. And then you have rational
00:24:38
economic agents etc. neuroscience right
00:24:42
neuroscience neurons as information
00:24:45
processing units. So we know that um if
00:24:49
you look at the way the brain works okay
00:24:51
so it's like a neural network basically
00:24:54
so there are neurons which are highly
00:24:56
interconnected with each other etc. So
00:24:59
neuroscience is a very important concept
00:25:01
there. Psychology, cognitive science.
00:25:04
Okay. How do people behave, perceive,
00:25:08
process cognitive information, represent
00:25:10
knowledge, etc. Right? So when we uh
00:25:13
take in things, okay, how much can we
00:25:16
store in our brain? Right? So how do we
00:25:18
carry out the knowledge representation?
00:25:21
How do we do the abstraction? Okay. So
00:25:24
we can see that uh some information is
00:25:27
stored in the brain, some information is
00:25:29
forgotten, you know all those type of
00:25:31
things. So that cognitive science and
00:25:34
psychology is very important
00:25:36
to if we um you know study that it'll
00:25:40
help us in AI also computer engineering.
00:25:44
So building faster computers we know
00:25:47
that as the days go by there have been
00:25:50
faster and faster computers. So AI
00:25:53
started in the 1950s and since then
00:25:55
there's been so much improvement. Okay,
00:25:58
the uh computers have becoming faster
00:26:01
and faster and the storage is is gone up
00:26:04
like anything, right? So all these
00:26:06
things actually help you in AI and um
00:26:10
you're able to get better and better
00:26:13
programs and and more uh accurate
00:26:15
programs because of this. Then control
00:26:18
theory.
00:26:21
So, so when you have an objective
00:26:24
function which you need to
00:26:26
uh maximize, you use a control theory,
00:26:30
feedback systems, etc. And linguistic,
00:26:33
how do you represent
00:26:35
uh knowledge? How do you uh work with
00:26:38
natural language? All these things.
00:26:40
Okay. So all these are something like um
00:26:44
the parent disciplines making it a
00:26:47
interdisciplinary area
00:26:50
basically. So the foundation of AI is
00:26:53
based on mathematics, neuroscience,
00:26:55
control theory, linguistics etc. Okay.
00:26:58
So mathematics is very important because
00:27:02
um um if you see uh it's only after say
00:27:06
1990 or so that AI picked up in all
00:27:10
these things okay we had things like uh
00:27:12
decision theory utility theory uh
00:27:16
mathematics um in speech recognition you
00:27:19
have uh mark of processes which
00:27:22
processes which came up etc. So you find
00:27:25
that uh it became a very sort of going
00:27:28
more and more towards
00:27:30
uh you know uh a full mathematical model
00:27:34
of the AI
00:27:37
systems. So in
00:27:39
mathematics okay logical methods are are
00:27:43
used. Okay. So when you use logical
00:27:45
methods obviously you have boolean logic
00:27:48
where you have everything represented by
00:27:51
zeros and ones and then you have fuzzy
00:27:53
logic where in fuzzy logic it can belong
00:27:57
to it can be a member of
00:27:59
um um it can be a member uh you know
00:28:04
membership with with a membership value
00:28:07
basically and then uncertainty obviously
00:28:11
um most of from the systems which we are
00:28:15
working on is based on uncertainty. So
00:28:19
when you have uncertainty you need to be
00:28:21
able to handle that uncertainty. Okay.
00:28:25
And this is done by using probability
00:28:27
theory other types of logics like model
00:28:30
temporal logics etc. You have
00:28:33
non-monotonic logic. So all these things
00:28:36
actually help you to deal with uh the
00:28:39
uncertainty. So
00:28:43
um neuroscience how the brain
00:28:47
works. Okay. How close are we to have a
00:28:50
mechanical
00:28:51
brain? Uh more recent studies use
00:28:53
accurate sensors to correlate brain
00:28:55
activity to human thought. Earlier
00:28:58
studies used injured and abnormal people
00:29:00
to understand which parts of the brain
00:29:02
work and which don't etc. Okay. So
00:29:05
neuroscience also goes into artificial
00:29:08
intelligence.
00:29:10
then control theory. So what you're
00:29:14
basically uh trying to do is uh you in
00:29:18
control theory the feedback that is used
00:29:21
right so you're trying to use that okay
00:29:26
uh and uh see how it can be used in AI
00:29:31
basically so linguistic speech
00:29:35
uh with speech is a is a big part in
00:29:39
human intelligence it's only through
00:29:41
speech that we understand things and we
00:29:43
use uh these uh you know what we
00:29:47
understand
00:29:49
um it it goes into our brain okay to
00:29:51
form human intelligence. So speech is a
00:29:54
very important
00:29:55
aspect. Um so so the uh language is
00:30:00
important there. For example, you can
00:30:02
see that children
00:30:04
uh they would have um you know as they
00:30:08
grow older, you find that they are able
00:30:10
to create sentences which they have
00:30:11
never heard of before, right? So um
00:30:16
language is is actually very important
00:30:19
and uh we we actually if you see we
00:30:22
think in a particular language for
00:30:24
example um for example I may think in
00:30:27
English right? So um the thinking and
00:30:31
language is actually sort of very
00:30:34
closely connected to each
00:30:37
other. Okay. So now what we are going to
00:30:41
uh talk about is something like the
00:30:42
history of AI. Okay. So uh we'll stop
00:30:47
here. Okay. And take a break and
00:30:50
continue this in the is in the next
00:30:52
session. Thank you.
