0
1
00:00:01,810 --> 00:00:07,540
Hello, in this session, we are going to talk about the history of artificial intelligence.
1

2
00:00:09,370 --> 00:00:13,870
Alan Turing introduced the Turing test in 1950s.
2

3
00:00:15,200 --> 00:00:20,050
Let's see what is the Alan Turing test? In those days,
3

4
00:00:20,080 --> 00:00:24,580
there were some vital questions like ''what is consciousness?''
4

5
00:00:24,850 --> 00:00:27,070
''Can an artificial machine really think?''
5

6
00:00:27,070 --> 00:00:31,570
"Does the mind just consist of neurons in the brain?"
6

7
00:00:31,870 --> 00:00:34,460
"Or is there something spark at its core?"
7

8
00:00:35,350 --> 00:00:42,280
For many, these have been vital consideration for the future of artificial intelligence.
8

9
00:00:42,560 --> 00:00:50,650
But British computer scientist Alan Turing decided to disregard all those questions in favour of a much
9

10
00:00:50,650 --> 00:00:51,530
simpler one.
10

11
00:00:52,000 --> 00:00:54,850
"Can a computer talk like a human?"
11

12
00:00:55,690 --> 00:01:03,460
This question led to an idea of measuring artificial intelligence that would famously come to be known
12

13
00:01:03,460 --> 00:01:04,840
as a Turing test.
13

14
00:01:05,440 --> 00:01:07,510
Let's take a look at these Turing test.
14

15
00:01:09,810 --> 00:01:19,740
In 1950, paper, computer, machinery, and intelligence Turing proposed the following game: Human Judge
15

16
00:01:19,740 --> 00:01:28,160
has a text conversation with unseen players and evaluates their responses to pass the test.
16

17
00:01:28,170 --> 00:01:33,780
A computer must be able to replace one of the players without changing the result.
17

18
00:01:34,260 --> 00:01:42,690
In other words, a computer would be considered intelligent if its conversation couldn't be easily distinguished
18

19
00:01:42,690 --> 00:01:45,660
from a human. In this test,
19

20
00:01:45,780 --> 00:01:51,920
if the machine is able to fool 70 percent of judges, then the machine passed the test.
20

21
00:01:52,650 --> 00:02:00,210
However, it wasn't a successful test in Turing's days, and he identified the problem.
21

22
00:02:00,210 --> 00:02:01,560
may be the memory.
22

23
00:02:01,980 --> 00:02:11,400
Turing predicted that by the year 2000, machines with 100 megabytes of memory would be able to easily
23

24
00:02:11,400 --> 00:02:12,590
pass this test.
24

25
00:02:12,960 --> 00:02:18,720
But we all know even today's machines have far more memory than that.
25

26
00:02:19,140 --> 00:02:26,760
Only a few of them have succeeded to pass this test and those which have succeeded, they just focused
26

27
00:02:26,760 --> 00:02:31,170
more on finding clever ways of fooling the judges.
27

28
00:02:31,650 --> 00:02:39,750
For instance, a machine which pretended that he is a 13 year old boy and not really good in English,
28

29
00:02:39,810 --> 00:02:44,460
was able to pass the test and fool the judges by 30 percent.
29

30
00:02:46,320 --> 00:02:53,810
John McCarthy in 1955 defines it as a science and engineering of making intelligent machines.
30

31
00:02:54,630 --> 00:03:01,890
It was the first time that someone was using those two words in one sentence intelligent and machines.
31

32
00:03:02,130 --> 00:03:08,440
He was calling machines intelligence, which was only used for human (this word), but now they just use that (intelligence)
32

33
00:03:08,440 --> 00:03:09,360
for machines.
33

34
00:03:09,900 --> 00:03:16,680
In 1950, a John McCarthy also introduced Lisp programming language, which was widely used in the
34

35
00:03:16,680 --> 00:03:20,250
field of artificial intelligence and programming machines.
35

36
00:03:21,090 --> 00:03:31,830
Prolog, which still is used, was introduced in 1972. Another important event of the history of artificial
36

37
00:03:31,830 --> 00:03:35,790
intelligence, we can mention about Deep Blue Project of IBM.
37

38
00:03:36,300 --> 00:03:40,960
Deep Blue defeated Garry Kasparov in 1997.
38

39
00:03:41,280 --> 00:03:48,390
Let's take a look at this event and see why this is very huge and important in the history of artificial
39

40
00:03:48,390 --> 00:03:49,000
intelligence.
40

41
00:03:50,160 --> 00:03:59,610
On May 11, 1997, an IBM computer called IBM Deep Blue beat the World Chess Champion.
41

42
00:04:00,120 --> 00:04:06,890
Their match lasts for several days and received massive media coverage all around the world.
42

43
00:04:07,380 --> 00:04:16,080
It was a classic plotline of man vs machines, and it was very huge because that was the first time
43

44
00:04:16,080 --> 00:04:19,610
that a machine was able to defeat a human.
44

45
00:04:19,980 --> 00:04:29,340
And the title of many newspapers in those days was The machines are taking over humans and the machines
45

46
00:04:29,340 --> 00:04:32,700
are winning over the human race and something like that.
46

47
00:04:32,970 --> 00:04:34,920
It was very huge on those days.
47

48
00:04:35,670 --> 00:04:40,500
Let's take a look at another important event on the history of artificial intelligence.
48

49
00:04:40,500 --> 00:04:49,590
We can mention about another project of IBM, which is Watson; IBM Build Watson in 2011, and won
49

50
00:04:49,590 --> 00:04:51,660
Jeopardy, which is a quiz show.
50

51
00:04:53,400 --> 00:05:00,990
Watson is a question answering computer system capable of answering questions posed in a natural language.
51

52
00:05:01,320 --> 00:05:09,690
Watson had access to two hundred million pages of structured and unstructured content, consuming four
52

53
00:05:09,690 --> 00:05:14,580
terabyte of storage, including the full text of Wikipedia.
53

54
00:05:14,820 --> 00:05:18,390
But of course, during the game it wasn't connected to the Internet.
54

55
00:05:19,260 --> 00:05:23,280
This computer was able to process the question and answer all the question.
55

56
00:05:23,280 --> 00:05:27,630
And one day, I guess, a million dollars prize for this show.
56

57
00:05:28,320 --> 00:05:33,930
Let's move on to another important event, which we can mention about Alpha Go zero!
57

58
00:05:34,530 --> 00:05:42,600
This is the most recent one, which is on 2017, Alpha Zero was taught how to play Go, but given no
58

59
00:05:42,600 --> 00:05:44,700
additional instruction.
59

60
00:05:45,030 --> 00:05:46,590
That is pretty amazing.
60

61
00:05:46,590 --> 00:05:47,880
Let's take a look at it.
61

62
00:05:48,810 --> 00:05:57,320
Go is known as the most challenging classical game for artificial intelligence because of its complexity.
62

63
00:05:58,710 --> 00:06:00,000
This is the game of Go.
63

64
00:06:00,010 --> 00:06:01,320
This is a Chinese game.
64

65
00:06:01,320 --> 00:06:09,230
Two players using either white or black stones take turns placing their stones on the board.
65

66
00:06:10,060 --> 00:06:19,330
The goal of this game is to surround and capture their opponent's stones or as strategically create
66

67
00:06:19,390 --> 00:06:28,090
spaces of territories. The rules of the game are pretty simple, but actually this is one of the most
67

68
00:06:28,090 --> 00:06:36,730
complex games because there are and an astonishing 10 to power of one hundred seventy possible board
68

69
00:06:36,730 --> 00:06:37,910
configurations.
69

70
00:06:38,200 --> 00:06:42,040
This is more than the number of atoms in the known universe.
70

71
00:06:42,430 --> 00:06:47,020
This makes the game Go and one of the most complex ones.
71

72
00:06:47,020 --> 00:06:54,190
And it's more complicated than chess because in a chess game we can predict how many possible moves
72

73
00:06:54,190 --> 00:06:56,980
are and we can just program it based on that.
73

74
00:06:57,130 --> 00:07:02,230
What programming and predicting the possible ways in this game is almost impossible.
74

75
00:07:02,830 --> 00:07:06,280
Now, let's take a look at these Alpha
75

76
00:07:06,280 --> 00:07:13,150
Go and see why this is very huge and very interesting in the field of artificial intelligence.
76

77
00:07:13,600 --> 00:07:20,470
Alpha Go Zero was taught how to play the game Go, but given no additional instruction.
77

78
00:07:20,590 --> 00:07:28,810
So how this is possible, no instruction?! instead. it learns the best moves over time simply by
78

79
00:07:28,810 --> 00:07:36,360
playing millions of games against itself. They just let the machine to explore the game.
79

80
00:07:36,460 --> 00:07:44,020
Just imagine when you start playing a new your game, sometimes you don't really go to read the instruction
80

81
00:07:44,020 --> 00:07:46,570
of the game to see how to play this game.
81

82
00:07:46,750 --> 00:07:48,850
You will just explore it by yourself.
82

83
00:07:48,850 --> 00:07:54,640
You will start maybe pressing on some buttons on the keyboard based on your experience sometimes.
83

84
00:07:54,880 --> 00:07:59,710
And you see, OK, so this 'F' button is for shooting!!
84

85
00:07:59,710 --> 00:08:06,460
You will just find it out yourself and then by a few times losing to the machine, you will realize
85

86
00:08:06,460 --> 00:08:14,140
that, OK, if I just shoot to these enemies, then I can have a scores and I can go to the next level.
86

87
00:08:14,620 --> 00:08:19,690
Basically you are learning based on your experience, you're experiencing it yourself.
87

88
00:08:20,050 --> 00:08:24,100
The thing that happened in this deep mine project is the same.
88

89
00:08:24,460 --> 00:08:31,840
They just let the machine to play against himself and learn the best strategy and the best way to
89

90
00:08:31,840 --> 00:08:33,430
predict and win the game.
90

91
00:08:35,080 --> 00:08:43,060
And this is a very huge step in a field of artificial intelligence, because this time the machine is
91

92
00:08:43,060 --> 00:08:48,460
finding new ways and a knowledge that wasn't taught by humans.
92

93
00:08:48,700 --> 00:08:51,760
And it's just learning it by itself.
93

94
00:08:52,120 --> 00:08:58,300
So that's why it's very important and it's a very huge step in thr field of artificial intelligence,
94

95
00:08:58,780 --> 00:09:00,640
not only because of this game,
95

96
00:09:00,910 --> 00:09:03,070
This is a technique that matters here.
96

97
00:09:03,070 --> 00:09:11,050
And later, we can use these techniques for solving some  complex problems in the field of Medicine and
97

98
00:09:11,050 --> 00:09:14,380
in the field of business strategies and different things.