Can a robot pass a university entrance exam? | Noriko Arai

205,788 views ใƒป 2017-09-13

TED


ืื ื ืœื—ืฅ ืคืขืžื™ื™ื ืขืœ ื”ื›ืชื•ื‘ื™ื•ืช ื‘ืื ื’ืœื™ืช ืœืžื˜ื” ื›ื“ื™ ืœื”ืคืขื™ืœ ืืช ื”ืกืจื˜ื•ืŸ.

ืžืชืจื’ื: Shlomo Adam ืžื‘ืงืจ: Sigal Tifferet
ื”ื™ื•ื ืื“ื‘ืจ ืขืœ ื”ืชื‘ื•ื ื” ื”ืžืœืื›ื•ืชื™ืช ืฉื‘ื—ื™ื™ื ื•.
00:13
Today, I'm going to talk about AI and us.
0
13014
3660
00:18
AI researchers have always said
1
18206
2143
ื›ืœ ื”ื—ื•ืงืจื™ื ืืžืจื• ืชืžื™ื“
00:20
that we humans do not need to worry,
2
20373
2594
ืฉืœื ื•, ื‘ื ื™ ื”ืื“ื, ืื™ืŸ ืžื” ืœื“ืื•ื’,
00:22
because only menial jobs will be taken over by machines.
3
22991
3580
ื›ื™ ื”ืžื›ื•ื ื•ืช ืชื›ื‘ื•ืฉื ื” ืจืง ืืช ื”ืขื‘ื•ื“ื•ืช ื”ืฉื—ื•ืจื•ืช.
00:27
Is that really true?
4
27274
1603
ื”ืื ื–ื” ื‘ืืžืช ื ื›ื•ืŸ?
00:30
They have also said that AI will create new jobs,
5
30365
3827
ื”ื ื’ื ืืžืจื• ืฉื”ืชื‘ื•ื ื” ื”ืžืœืื›ื•ืชื™ืช ืชื™ืฆื•ืจ ืžืงื•ืžื•ืช ืขื‘ื•ื“ื” ื—ื“ืฉื™ื
00:34
so those who lose their jobs will find a new one.
6
34216
3411
ื•ืžื™ ืฉื™ืื‘ื“ื• ืืช ืžืงื•ื ืขื‘ื•ื“ืชื, ื™ื•ื›ืœื• ืœืžืฆื•ื ืขื‘ื•ื“ื” ื—ื“ืฉื”.
00:38
Of course.
7
38264
1355
ื‘ื˜ื—.
00:39
But the real question is:
8
39643
2172
ืื‘ืœ ื”ืฉืืœื” ื”ืืžื™ืชื™ืช ื”ื™ื
00:41
How many of those who may lose their jobs to AI
9
41839
4105
ื›ืžื” ืžืืœื” ืฉื™ืคืกื™ื“ื• ืืช ืžืงื•ื ืขื‘ื•ื“ืชื ืœืชื‘ื•ื ื” ื”ืžืœืื›ื•ืชื™ืช
00:45
will be able to land a new one,
10
45968
2489
ื™ื•ื›ืœื• ืœืžืฆื•ื ืขื‘ื•ื“ื” ื—ื“ืฉื”,
00:48
especially when AI is smart enough to learn better than most of us?
11
48481
5850
ื‘ืžื™ื•ื—ื“ ื›ืฉื”ืชื‘ื•ื ื” ื”ืžืœืื›ื•ืชื™ืช ื—ื›ืžื” ืžืกืคื™ืง ื›ื“ื™ ืœืœืžื•ื“ ื˜ื•ื‘ ื™ื•ืชืจ ืžืžืจื‘ื™ืชื ื•?
00:55
Let me ask you a question:
12
55397
2185
ื”ื‘ื” ื•ืืฉืืœ ืืชื›ื ืฉืืœื”:
00:58
How many of you think
13
58666
1798
ื›ืžื” ืžื›ื ืกื‘ื•ืจื™ื
01:00
that AI will pass the entrance examination of a top university by 2020?
14
60488
6094
ืฉื”ืชื‘ื•ื ื” ื”ืžืœืื›ื•ืชื™ืช ืชื•ื›ืœ ืœืขื‘ื•ืจ ื‘ื—ื™ื ืช ื›ื ื™ืกื”
ืœืื•ื ื™ื‘ืจืกื™ื˜ื” ืžื•ื‘ื—ืจืช ืขื“ ืฉื ืช 2020?
01:07
Oh, so many. OK.
15
67836
2457
ืจื‘ื™ื ืžื›ื. ื‘ืกื“ืจ.
01:10
So some of you may say, "Of course, yes!"
16
70317
4358
ืื– ื—ืœืง ืžื›ื ืื•ืžืจื™ื, "ื›ืžื•ื‘ืŸ ืฉื›ืŸ!
01:15
Now singularity is the issue.
17
75369
2134
"ืขื›ืฉื™ื• ื”ืกื™ื ื’ื•ืœืจื™ื•ืช ื”ื™ื ื”-ื“ื‘ืจ!",
01:18
And some others may say, "Maybe,
18
78590
3095
ื•ืื—ืจื™ื ืื•ืœื™ ืื•ืžืจื™ื, "ืื•ืœื™,
01:21
because AI already won against a top Go player."
19
81709
4508
"ื›ื™ ื”ืชื‘ื•ื ื” ื”ืžืœืื›ื•ืชื™ืช ื›ื‘ืจ ื ื™ืฆื—ื” ืฉื—ืงืŸ ื’ื• ืžืขื•ืœื”"
01:27
And others may say, "No, never. Uh-uh."
20
87213
3681
ื•ืื™ืœื• ืื—ืจื™ื ืื•ืžืจื™ื, "ืœื. ืื™ืŸ ืžืฆื‘"
01:32
That means we do not know the answer yet, right?
21
92195
3593
ื›ืœื•ืžืจ, ืขื“ื™ื™ืŸ ืื™ื ื ื• ื™ื•ื“ืขื™ื ืžื”ื™ ื”ืชืฉื•ื‘ื”.
01:36
So that was the reason why I started Todai Robot Project,
22
96268
4960
ืžืกื™ื‘ื” ื–ื• ื”ืชื—ืœืชื™ ื‘ืžื™ื–ื ื”ืจื•ื‘ื•ื˜ื™ืงื” ืฉืœ ื˜ื•ื“ืื™,
01:41
making an AI which passes the entrance examination
23
101252
3872
ื‘ืžื˜ืจื” ืœื™ืฆื•ืจ ืชื‘ื•ื ื” ืžืœืื›ื•ืชื™ืช ืฉืชืขื‘ื•ืจ ื‘ื—ื™ื ืช ื›ื ื™ืกื”
01:45
of the University of Tokyo,
24
105148
2589
ืœืื•ื ื™ื‘ืจืกื™ื˜ืช ื˜ื•ืงื™ื•.
01:47
the top university in Japan.
25
107761
2537
ื”ืื•ื ื™ื‘ืจืกื™ื˜ื” ื”ื›ื™ ื˜ื•ื‘ื” ื‘ื™ืคืŸ.
01:51
This is our Todai Robot.
26
111464
2548
ื–ื”ื• ืจื•ื‘ื•ื˜ ื˜ื•ื“ืื™ ืฉืœื ื•.
ื•ืžื•ื—ื• ืฉืœ ื”ืจื•ื‘ื•ื˜ ืคื•ืขืœ ื›ืžื•ื‘ืŸ ื‘ืฉืจืช ืžืจื•ื—ืง
01:56
And, of course, the brain of the robot is working in the remote server.
27
116131
5810
02:02
It is now writing a 600-word essay
28
122747
4222
ื›ืจื’ืข ืจื•ืื™ื ืื•ืชื• ื›ื•ืชื‘ ื—ื™ื‘ื•ืจ ื‘ืŸ 600 ืžื™ืœื”
02:06
on maritime trade in the 17th century.
29
126993
4119
ืขืœ ื”ืกื—ืจ ื”ื™ืžื™ ื‘ืžืื” ื”-17
ืžื” ื“ืขืชื›ื ืขืœ ื–ื”?
02:11
How does that sound?
30
131136
1765
ื•ืžื“ื•ืข ืœืงื—ืชื™ ืืช ื‘ื—ื™ื ืช ื”ื›ื ื™ืกื” ื›ืžื“ื“ ืœื‘ื™ืฆื•ืขื™ื ืฉืœื•?
02:14
Why did I take the entrance exam as its benchmark?
31
134113
4104
ื›ื™ ื—ืฉื‘ืชื™ ืฉืขืœื™ื ื• ืœื—ืงื•ืจ ืืช ื‘ื™ืฆื•ืขื™ื” ืฉืœ ื”ืชื‘ื•ื ื” ื”ืžืœืื›ื•ืชื™ืช
02:19
Because I thought we had to study the performance of AI
32
139098
4741
02:23
in comparison to humans,
33
143863
2114
ื‘ื”ืฉื•ื•ืื” ืœื‘ื™ืฆื•ืขื™ื ืื ื•ืฉื™ื™ื
ื‘ืžื™ื•ื—ื“ ื‘ื”ื™ืงืคื™ื ื•ื‘ืชื—ื•ืžื™ ืžื•ืžื—ื™ื•ืช
02:26
especially on the skills and expertise
34
146001
2860
02:28
which are believed to be acquired only by humans
35
148885
4088
ืฉืœืคื™ ื”ืืžื•ื ื”, ืจืง ื‘ื ื™-ืื“ื ื™ื›ื•ืœื™ื ืœืจื›ื•ืฉ
02:32
and only through education.
36
152997
2335
ื•ืจืง ื‘ืืžืฆืขื•ืช ื”ืฉื›ืœื”.
02:35
To enter Todai, the University of Tokyo,
37
155782
4043
ื›ื“ื™ ืœื”ื™ื›ื ืก ืœื˜ื•ื“ืื™, ืื•ื ื™ื‘ืจืกื™ื˜ืช ื˜ื•ืงื™ื•,
02:39
you have to pass two different types of exams.
38
159849
4421
ื™ืฉ ืœืขื‘ื•ืจ ืฉื ื™ ืกื•ื’ื™ ื‘ื—ื™ื ื•ืช.
02:44
The first one is a national standardized test
39
164294
3760
ื”ืจืืฉื•ื ื” ื”ื™ื ืžื‘ื—ืŸ ืœืื•ืžื™ ืžืชื•ืงื ืŸ
ื‘ืกื’ื ื•ืŸ ืืžืจื™ืงืื™.
02:48
in multiple-choice style.
40
168078
2403
02:50
You have to take seven subjects
41
170505
2455
ืฆืจื™ืš ืœื‘ื—ื•ืจ ืฉื‘ืขื” ื ื•ืฉืื™ื
02:52
and achieve a high score --
42
172984
1955
ื•ืœื”ืฉื™ื’ ืฆื™ื•ืŸ ื’ื‘ื•ื” --
02:54
I would say like an 85 percent or more accuracy rate --
43
174963
4772
ื”ื™ื™ืชื™ ืื•ืžืจืช 84% ืื• ืื—ื•ื– ื“ื™ื•ืง ื’ื‘ื•ื” ื™ื•ืชืจ --
02:59
to be allowed to take the second stage written test
44
179759
4087
ื›ื“ื™ ืœื’ืฉืช ืœืฉืœื‘ ื”ืฉื ื™ ืฉืœ ื”ื‘ื—ื™ื ื” ื‘ื›ืชื‘
03:03
prepared by Todai.
45
183870
2159
ืฉื ื›ืชื‘ ื‘ื˜ื•ื“ืื™.
03:06
So let me first explain how modern AI works,
46
186994
5317
ืื– ืชื—ื™ืœื” ืืกื‘ื™ืจ ืื™ืš ืขื•ื‘ื“ืช ื”ืชื‘ื•ื ื” ื”ืžืœืื›ื•ืชื™ืช ื”ืžื•ื“ืจื ื™ืช,
03:12
taking the "Jeopardy!" challenge as an example.
47
192335
3069
ื•ืืžื—ื™ืฉ ื–ืืช ื‘ืขื–ืจืช ืžืฉื—ืง ื”ื˜ืจื™ื•ื•ื™ื” "ื’'ืคืจื“ื™".
03:17
Here is a typical "Jeopardy!" question:
48
197539
3079
ื”ื ื” ืฉืืœืช "ื’'ืคืจื“ื™" ืื•ืคื™ื™ื ื™ืช:
03:20
"Mozart's last symphony shares its name with this planet."
49
200642
4461
"ืฉื ื”ืกื™ืžืคื•ื ื™ื” ื”ืื—ืจื•ื ื” ืฉืœ ืžื•ืฆืจื˜ ื–ื”ื” ืœืฉืžื• ืฉืœ ื›ื•ื›ื‘-ืœื›ืช ื–ื”."
03:26
Interestingly, a "Jeopardy!" question always asks,
50
206195
4013
ืžืขื ื™ื™ืŸ ืฉืฉืืœื•ืช "ื’'ืคืจื“ื™"
ืžืกืชื™ื™ืžื•ืช ืชืžื™ื“ ื‘"ื–ื”" ืื• "ื–ื•":
03:30
always ends with "this" something:
51
210232
3328
03:33
"this" planet, "this" country,
52
213584
2827
"ื›ื•ื›ื‘-ืœื›ืช ื–ื”", "ืืจืฅ ื–ื•",
03:36
"this" rock musician, and so on.
53
216435
2608
"ืžื•ืกื™ืงืื™ ื”ืจื•ืง ื”ื–ื”", ื•ื›ืŸ ื”ืœืื”.
ื‘ืžืœื™ื ืื—ืจื•ืช, ื‘"ื’'ืคืจื“ื™" ืื™ืŸ ืกื•ื’ื™ื ืจื‘ื™ื ืฉืœ ืฉืืœื•ืช,
03:39
In other words, "Jeopardy!" doesn't ask many different types of questions,
54
219067
4299
03:43
but a single type,
55
223390
1837
ืืœื ืกื•ื’ ืื—ื“,
03:45
which we call "factoid questions."
56
225251
2536
ื”ื ืงืจื "ืขื•ื‘ื“ื” ืœืžื—ืฆื”".
03:48
By the way, do you know the answer?
57
228975
2167
ืื’ื‘, ืืชื ื™ื•ื“ืขื™ื ืžื”ื™ ื”ืชืฉื•ื‘ื”?
03:53
If you do not know the answer and if you want to know the answer,
58
233980
4055
ืžื™ ืฉืื™ื ื• ื™ื•ื“ืข ืืช ื”ืชืฉื•ื‘ื” ื•ืจื•ืฆื” ืœืœืžื•ื“ ืื•ืชื”,
ืžื” ื”ื•ื ืขื•ืฉื”?
03:58
what would you do?
59
238059
1287
ืžื—ืคืฉ ื‘"ื’ื•ื’ืœ", ื›ืžื•ื‘ืŸ.
04:00
You Google, right? Of course.
60
240160
3132
04:03
Why not?
61
243316
1480
ืœืžื” ืœื?
04:04
But you have to pick appropriate keywords
62
244820
3592
ืื‘ืœ ืขืœื™ื›ื ืœื‘ื—ื•ืจ ืืช ืžื™ืœื•ืช ื”ืžืคืชื— ื”ื ื›ื•ื ื•ืช,
04:08
like "Mozart," "last" and "symphony" to search.
63
248436
4364
ืœืžืฉืœ: "ืžื•ืฆืจื˜", "ืื—ืจื•ื ื”" ื•"ืกื™ืžืคื•ื ื™ื”" ื›ื“ื™ ืœื—ืคืฉ.
04:13
The machine basically does the same.
64
253462
2400
ื”ืžื›ื•ื ื”, ื‘ืขืงืจื•ืŸ, ืขื•ืฉื” ืื•ืชื• ื“ื‘ืจ,
04:16
Then this Wikipedia page will be ranked top.
65
256457
4660
ื•ืื– ื“ืฃ ื”"ื•ื•ื™ืงื™ืคื“ื™ื”" ื”ื–ื” ืžื“ื•ืจื’ ืจืืฉื•ืŸ,
04:21
Then the machine reads the page.
66
261840
1908
ื•ื”ืžื›ื•ื ื” ืงื•ืจืืช ืืช ื”ื“ืฃ.
04:23
No, uh-uh.
67
263772
1171
ืœื. ืžืžืฉ ืœื.
04:25
Unfortunately, none of the modern AIs,
68
265470
3462
ืœืฆืขืจื ื•, ืื™ืŸ ืชื‘ื•ื ื” ืžืœืื›ื•ืชื™ืช ืžื•ื“ืจื ื™ืช,
04:28
including Watson, Siri and Todai Robot,
69
268956
3968
ื›ื•ืœืœ "ื•ื•ื˜ืกื•ืŸ", "ืกื™ืจื™" ื•ื”ืจื•ื‘ื•ื˜ ืฉืœ ื˜ื•ื“ืื™,
04:32
is able to read.
70
272948
1661
ืฉื™ื•ื“ืขืช ืœืงืจื•ื.
04:35
But they are very good at searching and optimizing.
71
275437
3800
ืื‘ืœ ื”ืŸ ื™ื•ื“ืขื•ืช ื”ื™ื˜ื‘ ืœื—ืคืฉ ื•ืœืžื˜ื‘ ืืช ื”ื—ื™ืคื•ืฉ.
ื”ืžื›ื•ื ื” ืชื–ื”ื”
04:40
It will recognize
72
280158
2023
04:42
that the keywords "Mozart," "last" and "symphony"
73
282866
2935
ืฉืžื™ืœื•ืช ื”ืžืคืชื— "ืžื•ืฆืจื˜", "ืื—ืจื•ื ื”" ื•"ืกื™ืžืคื•ื ื™ื”"
04:45
are appearing heavily around here.
74
285825
2903
ืžืจื‘ื•ืช ืœื”ื•ืคื™ืข ื‘ืื–ื•ืจ ื”ื–ื”.
04:49
So if it can find a word which is a planet
75
289790
4375
ื›ืš ืฉื”ื™ื ืžืกื•ื’ืœืช ืœืžืฆื•ื ืฉื ืฉืœ ื›ื•ื›ื‘-ืœื›ืช
04:54
and which is co-occurring with these keywords,
76
294189
3648
ืฉืžื•ืคื™ืข ื’ื ื”ื•ื ื‘ืžืงื‘ื™ืœ ืœืžื™ืœื•ืช ื”ืžืคืชื— ื”ืืœื”,
04:57
that must be the answer.
77
297861
1989
ื•ื”ื•ื ื‘ื•ื•ื“ืื™ ื”ืชืฉื•ื‘ื”.
05:00
This is how Watson finds the answer "Jupiter," in this case.
78
300762
5186
ื›ืš ื”ืฆืœื™ื— "ื•ื•ื˜ืกื•ืŸ" ืœืžืฆื•ื ืืช ื”ืชืฉื•ื‘ื” ื”ื ื›ื•ื ื”:
"ื™ื•ืคื™ื˜ืจ", ื‘ืžืงืจื” ื–ื”.
05:08
Our Todai Robot works similarly, but a bit smarter
79
308433
4049
ืจื•ื‘ื•ื˜ ื˜ื•ื“ืื™ ืฉืœื ื• ืคื•ืขืœ ื‘ื“ื•ืžื” ืœื›ืš, ืื‘ืœ ื‘ืฆื•ืจื” ืงืฆืช ื™ื•ืชืจ ื—ื›ืžื”.
05:12
in answering history yes-no questions,
80
312506
3239
ื›ืฉื”ื•ื ืžืฉื™ื‘ ืœืฉืืœื•ืช ื›ืŸ-ืœื ื‘ื”ื™ืกื˜ื•ืจื™ื” ,
05:16
like, "'Charlemagne repelled the Magyars.' Is this sentence true or false?"
81
316560
5663
ื›ืžื•, "ืงืจืœ ื”ื’ื“ื•ืœ ื”ื“ืฃ ืืช ื”ืžื’ื™ืืจื™ื, ื ื›ื•ืŸ ืื• ืœื?"
05:23
Our robot starts producing a factoid question,
82
323181
4073
ื”ืจื•ื‘ื•ื˜ ืฉืœื ื• ืžืชื—ื™ืœ ืœื™ื™ืฆืจ ืฉืืœื•ืช ืขื•ื‘ื“ื”-ืœืžื—ืฆื”,
05:27
like: "Charlemagne repelled [this person type]" by itself.
83
327278
4899
ื›ืžื•: "ืงืจืœ ื”ื’ื“ื•ืœ ื”ื“ืฃ [ืกื•ื’ ื”ืื“ื ื”ื–ื”]" ื‘ืขืฆืžื•
05:32
Then, "Avars" but not "Magyars" is ranked top.
84
332995
4732
ืœืื—ืจ ืžื›ืŸ, "ืžื’ื™ืืจื™ื" ืื™ื ื” ืžื“ื•ืจื’ืช ื‘ืจืืฉ.
05:38
This sentence is likely to be false.
85
338357
3049
ื”ืžืฉืคื˜ ื”ื–ื” ื›ื ืจืื” ื›ื•ื–ื‘.
05:42
Our robot does not read, does not understand,
86
342772
4860
ื”ืจื•ื‘ื•ื˜ ืฉืœื ื• ืื™ื ื• ืงื•ืจื ื•ืื™ื ื• ืžื‘ื™ืŸ,
05:48
but it is statistically correct in many cases.
87
348335
4144
ืืš ื‘ืื•ืคืŸ ืกื˜ื˜ื™ืกื˜ื™, ื”ื•ื ืฆื•ื“ืง ื‘ืžืงืจื™ื ืจื‘ื™ื.
ื‘ืฉืœื‘ ื”ืฉื ื™ ืฉืœ ื”ืžื‘ื—ืŸ ื‘ื›ืชื‘
05:54
For the second stage written test,
88
354147
2508
05:56
it is required to write a 600-word essay like this one:
89
356679
5106
ื”ื•ื ื ื“ืจืฉ ืœื›ืชื•ื‘ ื—ื™ื‘ื•ืจ ื‘ืŸ 600 ืžื™ืœื”, ื›ื–ื”:
06:01
[Discuss the rise and fall of the maritime trade
90
361809
2278
[ื ืชื— ืืช ืขืœื™ื™ืชื• ื•ื™ืจื™ื“ืชื• ืฉืœ ื”ืกื—ืจ ื”ื™ืžื™
ื‘ืžื–ืจื— ื•ื‘ื“ืจื•ื-ืžื–ืจื— ืืกื™ื” ื‘ืžืื” ื”-17],
06:04
in East and Southeast Asia in the 17th century ...]
91
364111
2422
06:06
and as I have shown earlier,
92
366557
1387
ื•ื›ืคื™ ืฉื”ืจืื™ืชื™ ืงื•ื“ื,
06:07
our robot took the sentences from the textbooks and Wikipedia,
93
367968
4194
ื”ืจื•ื‘ื•ื˜ ืฉืœื ื• ืœืงื— ืืช ื”ืžืฉืคื˜ื™ื ืžืกืคืจื™ ื”ืœื™ืžื•ื“ ื•ืž"ื•ื™ืงื™ืคื“ื™ื”",
06:12
combined them together,
94
372186
1961
ืฉื™ืœื‘ ืื•ืชื,
ื•ืžื™ื˜ื‘ ืื•ืชื ืœื›ืœืœ ื—ื™ื‘ื•ืจ
06:14
and optimized it to produce an essay
95
374171
3619
06:17
without understanding a thing.
96
377814
2207
ืžื‘ืœื™ ืœื”ื‘ื™ืŸ ื“ื‘ืจ.
(ืฆื—ื•ืง)
06:20
(Laughter)
97
380045
1737
06:21
But surprisingly, it wrote a better essay
98
381806
4895
ืืš ืœืžืจื‘ื” ื”ืคืœื, ื”ื•ื ื›ืชื‘ ื—ื™ื‘ื•ืจ ื˜ื•ื‘ ื™ื•ืชืจ
06:26
than most of the students.
99
386725
1561
ืžืจื•ื‘ ื”ืกื˜ื•ื“ื ื˜ื™ื.
06:28
(Laughter)
100
388310
2391
(ืฆื—ื•ืง)
06:30
How about mathematics?
101
390725
1529
ื•ืžื” ื‘ืžืชืžื˜ื™ืงื”?
06:33
A fully automatic math-solving machine
102
393354
3158
ืžื›ื•ื ื” ืื•ื˜ื•ืžื˜ื™ืช ืœื’ืžืจื™ ืœืคืชืจื•ืŸ ื‘ืขื™ื•ืช ืžืชืžื˜ื™ื•ืช
06:36
has been a dream
103
396536
1631
ื”ื™ืชื” ื‘ื’ื“ืจ ื—ืœื•ื
06:38
since the birth of the word "artificial intelligence,"
104
398191
4679
ืžืื– ื”ื•ืœื“ืช ื”ืžื•ื ื— "ืชื‘ื•ื ื” ืžืœืื›ื•ืชื™ืช",
06:43
but it has stayed at the level of arithmetic for a long, long time.
105
403785
6007
ืื‘ืœ ื‘ืžืฉืš ื–ืžืŸ ืจื‘ ืžืื“ ื–ื” ื ื•ืชืจ ื‘ืจืžืช ื”ื—ืฉื‘ื•ืŸ ื”ืคืฉื•ื˜.
06:51
Last year, we finally succeeded in developing a system
106
411530
5350
ื‘ืฉื ื” ืฉืขื‘ืจื” ื”ืฆืœื—ื ื• ืกื•ืฃ-ืกื•ืฃ ืœืคืชื— ืžืขืจื›ืช
06:56
which solved pre-university-level problems from end to end,
107
416904
5173
ืฉืคื•ืชืจืช ื‘ืื•ืคืŸ ืžื•ืฉืœื ื‘ืขื™ื•ืช ืงื“ื-ืื•ื ื™ื‘ืจืกื™ื˜ืื™ื•ืช,
ื›ืžื• ื–ืืช.
07:02
like this one.
108
422101
1262
[ืฉืืœื” ื‘ื’ื™ืื•ืžื˜ืจื™ื”]
07:05
This is the original problem written in Japanese,
109
425648
4002
ื–ืืช ื”ื‘ืขื™ื” ื”ืžืงื•ืจื™ืช, ื‘ื™ืคื ื™ืช,
07:09
and we had to teach it 2,000 mathematical axioms
110
429674
4397
ื•ื”ื™ื™ื ื• ืฆืจื™ื›ื™ื ืœืœืžื“ ืืช ื”ืžื›ื•ื ื” 2,000 ืืงืกื™ื•ืžื•ืช ืžืชืžื˜ื™ื•ืช
ื•-8,000 ืžืœื™ื ื‘ื™ืคื ื™ืช
07:14
and 8,000 Japanese words
111
434095
2774
07:16
to make it accept the problems written in natural language.
112
436893
4558
ื›ื“ื™ ืฉื”ื™ื ืชื•ื›ืœ ืœืงื‘ืœ ื‘ืขื™ื•ืช ืฉื›ืชื•ื‘ื•ืช ื‘ืฉืคื” ื˜ื‘ืขื™ืช.
07:22
And it is now translating the original problems
113
442234
3542
ื•ื›ืืŸ ื”ื™ื ืžืชืจื’ืžืช ืืช ื”ื‘ืขื™ื•ืช ื”ืžืงื•ืจื™ื•ืช
07:25
into machine-readable formulas.
114
445800
3139
ืœื ื•ืกื—ืื•ืช ืงืจื™ืื•ืช ืข"ื™ ืžื›ื•ื ื”.
07:30
Weird, but it is now ready to solve it, I think.
115
450578
6099
ืžื•ื–ืจ, ืื‘ืœ ืขื›ืฉื™ื• ื”ื™ื ืžื•ื›ื ื” ืœืคืชื•ืจ ืื•ืชืŸ, ื ืจืื” ืœื™.
07:36
Go and solve it.
116
456701
1411
ืงื“ื™ืžื”, ืคืชืจื™ ืื•ืชืŸ.
07:38
Yes! It is now executing symbolic computation.
117
458818
4284
ื™ืฉ! ืขื›ืฉื™ื• ื”ื™ื ืžื‘ืฆืขืช ื—ื™ืฉื•ื‘ ืกื™ืžื‘ื•ืœื™.
ืžื•ื–ืจ ืขื•ื“ ื™ื•ืชืจ,
07:44
Even more weird,
118
464030
1580
07:45
but probably this is the most fun part for the machine.
119
465634
4825
ืื‘ืœ ืื•ืœื™ ื–ื” ื”ื—ืœืง ื”ื›ื™ ื›ื™ืคื™ ื‘ืฉื‘ื™ืœ ื”ืžื›ื•ื ื”.
07:50
(Laughter)
120
470483
2351
(ืฆื—ื•ืง)
07:52
Now it outputs a perfect answer,
121
472858
2815
ืขื›ืฉื™ื• ื”ื™ื ืžืคื™ืงื” ืชืฉื•ื‘ื” ืžื•ืฉืœืžืช,
07:55
though its proof is impossible to read, even for mathematicians.
122
475697
4707
ืœืžืจื•ืช ืฉื”ื”ื•ื›ื—ื” ืื™ื ื” ืงืจื™ืื”, ืืคื™ืœื• ืœืžืชืžื˜ื™ืงืื™ื.
08:02
Anyway, last year our robot was among the top one percent
123
482773
6961
ื‘ื›ืœ ืื•ืคืŸ, ื‘ืฉื ื” ืฉืขื‘ืจื” ื”ืจื•ื‘ื•ื˜ ืฉืœื ื• ื”ื™ื” ื‘ืื—ื•ื–ื•ืŸ ื”ืขืœื™ื•ืŸ
08:10
in the second stage written exam in mathematics.
124
490199
3633
ื‘ืฉืœื‘ 2 ืฉืœ ื”ืžื‘ื—ืŸ ื‘ื›ืชื‘ ื‘ืžืชืžื˜ื™ืงื”.
08:14
(Applause)
125
494652
3210
(ืžื—ื™ืื•ืช ื›ืคื™ื™ื)
08:18
Thank you.
126
498412
1311
ืชื•ื“ื”.
08:19
So, did it enter Todai?
127
499747
2471
ืื– ื”ืื ื”ื•ื ื”ืชืงื‘ืœ ืœื˜ื•ื“ืื™?
08:22
No, not as I expected.
128
502981
3058
ืœื, ืœื ื›ืคื™ ืฉืฆื™ืคื™ืชื™.
08:26
Why?
129
506783
1399
ื•ืžื“ื•ืข?
08:28
Because it doesn't understand any meaning.
130
508206
2639
ื›ื™ ื”ื•ื ืœื ืžื‘ื™ืŸ ืžืฉืžืขื•ื™ื•ืช.
ืืจืื” ืœื›ื ืฉื’ื™ืื” ืื•ืคื™ื™ื ื™ืช ืฉื”ื•ื ืขืฉื” ื‘ืžื‘ื—ืŸ ื‘ืื ื’ืœื™ืช.
08:32
Let me show you a typical error it made in the English test.
131
512308
4079
08:36
[Nate: We're almost at the bookstore. Just a few more minutes.
132
516411
2977
[-ื›ืžืขื˜ ื”ื’ืขื ื• ืœื—ื ื•ืช, -ืจื’ืข -- (?)
ืฉื ื™ ืื ืฉื™ื ืžื“ื‘ืจื™ื...
08:39
Sunil: Wait. ______ . Nate: Thank you! That always happens ...]
133
519412
3039
[-ื–ื” ืงื•ืจื” ืชืžื™ื“... -ืœื ืงืฉืจืช ืืช ื”ืฉืจื•ื›ื™ื?]
08:42
Two people are talking.
134
522475
1151
ืœื ื•, ืฉื™ื›ื•ืœื™ื ืœื”ื‘ื™ืŸ ืืช ื”ืกื™ื˜ื•ืืฆื™ื”,
08:43
For us, who can understand the situation --
135
523650
2054
08:45
[1. "We walked for a long time." 2. "We're almost there."
136
525704
2773
ื‘ืจื•ืจ ืฉืชืฉื•ื‘ื” ืžืก' 4 ื”ื™ื ื”ื ื›ื•ื ื”, ื ื›ื•ืŸ?
08:48
3. "Your shoes look expensive." 4. "Your shoelace is untied."]
137
528501
3032
[4. ื”ืฉืจื•ื›ื™ื ืฉืœืš ืคืจื•ืžื™ื]
08:51
it is obvious number four is the correct answer, right?
138
531557
2873
ืื‘ืœ ื”ืจื•ื‘ื•ื˜ ืฉืœ ื˜ื•ื“ืื™ ื‘ื—ืจ ื‘ืชืฉื•ื‘ื” ืžืก' 2,
08:54
But Todai Robot chose number two,
139
534454
2238
[2. ื›ืžืขื˜ ื”ื’ืขื ื•]
08:56
even after learning 15 billion English sentences
140
536716
5360
ืืคื™ืœื• ืื—ืจื™ ืฉืœืžื“ 15 ืžื™ืœื™ืืจื“ ืžืฉืคื˜ื™ื ื‘ืื ื’ืœื™ืช
ื‘ื˜ื›ื ื•ืœื•ื’ื™ื•ืช ืœืžื™ื“ื” ืขืžื•ืงื”.
09:02
using deep learning technologies.
141
542100
2728
09:07
OK, so now you might understand what I said:
142
547600
4172
ืื– ืขื›ืฉื™ื• ืื•ืœื™ ืชื‘ื™ื ื• ืืช ืžื” ืฉืืžืจืชื™:
09:12
modern AIs do not read,
143
552399
2648
ืชื‘ื•ื ื•ืช ืžืœืื›ื•ืชื™ื•ืช ืžื•ื“ืจื ื™ื•ืช ืœื ื™ื•ื“ืขื•ืช ืœืงืจื•ื ื•ืœื”ื‘ื™ืŸ.
09:15
do not understand.
144
555071
1413
09:17
They only disguise as if they do.
145
557516
3169
ื–ื” ืจืง ื ืจืื” ื›ืš.
09:24
This is the distribution graph
146
564867
2981
ื–ื”ื• ื’ืจืฃ ื”ื”ืชืคืœื’ื•ืช
09:27
of half a million students who took the same exam as Todai Robot.
147
567872
5777
ืฉืœ ื—ืฆื™ ืžื™ืœื™ื•ืŸ ืกื˜ื•ื“ื ื˜ื™ื ืฉื ื‘ื—ื ื• ื‘ืžื‘ื—ืŸ ืฉืจื•ื‘ื•ื˜ ื˜ื•ื“ืื™ ืขืฉื”.
09:34
Now our Todai Robot is among the top 20 percent,
148
574558
5165
ืจื•ื‘ื•ื˜ ื˜ื•ื“ืื™ ืฉืœื ื• ื ืžืฆื ื‘ื™ืŸ 20 ื”ืื—ื•ื–ื™ื ื”ืจืืฉื•ื ื™ื
09:40
and it was capable to pass
149
580986
2415
ื•ื”ื•ื ื”ืฆืœื™ื— ืœืขื‘ื•ืจ ืืช ืžื‘ื—ื ื™ ื”ืงื‘ืœื”
09:43
more than 60 percent of the universities in Japan --
150
583425
3941
ื‘ื™ื•ืชืจ ืž-60% ืžื”ืื•ื ื™ื‘ืจืกื™ื˜ืื•ืช ื‘ื™ืคืŸ
09:47
but not Todai.
151
587390
1377
ืืš ืœื ื‘ื˜ื•ื“ืื™.
ืื‘ืœ ืจืื• ืื™ืš ื”ื•ื ืขื‘ืจ ืืช ืžืจื‘ื™ืช
09:50
But see how it is beyond the volume zone
152
590116
4025
ืขื•ื‘ื“ื™-ื”ืฆื•ื•ืืจื•ืŸ-ื”ืœื‘ืŸ ื”ืขืชื™ื“ื™ื™ื.
09:54
of to-be white-collar workers.
153
594165
2864
ืื•ืœื™ ื ืจืื” ืœื›ื ืฉื”ื™ื™ืชื™ ืžืจื•ืฆื”.
10:00
You might think I was delighted.
154
600060
2858
10:03
After all, my robot was surpassing students everywhere.
155
603939
3971
ื”ืจื™ ื”ืจื•ื‘ื•ื˜ ืฉืœื™ ืขืœื” ืขืœ ื”ืžื•ืŸ ืกื˜ื•ื“ื ื˜ื™ื.
ืื‘ืœ ื–ื” ื–ื™ืขื–ืข ืื•ืชื™.
10:09
Instead, I was alarmed.
156
609022
2691
ืื™ืš ื™ื™ืชื›ืŸ ืฉื”ืžื›ื•ื ื” ื—ืกืจืช ื”ืชื‘ื•ื ื” ื”ื–ืืช ื’ื•ื‘ืจืช ื‘ื‘ื™ืฆื•ืขื™ื” ืขืœ ืกื˜ื•ื“ื ื˜ื™ื --
10:13
How on earth could this unintelligent machine outperform students --
157
613086
5607
10:18
our children?
158
618717
1292
ืขืœ ื”ื™ืœื“ื™ื ืฉืœื ื•, ื›ืŸ?
10:20
Right?
159
620033
1153
ื”ื—ืœื˜ืชื™ ืœื—ืงื•ืจ ืžื” ืงื•ืจื” ื‘ืขื•ืœื ื”ืื ื•ืฉื™.
10:22
I decided to investigate what was going on in the human world.
160
622101
4402
10:28
I took hundreds of sentences from high school textbooks
161
628542
4729
ืœืงื—ืชื™ ืžืื•ืช ืžืฉืคื˜ื™ื ืžืกืคืจื™ ืœื™ืžื•ื“ ืชื™ื›ื•ื ื™ื™ื
10:33
and made easy multiple-choice quizzes,
162
633859
3313
ื”ื›ื ืชื™ ืžื‘ื—ื ื™ื ืืžืจื™ืงืื™ื™ื ืงืœื™ื,
ื•ื‘ื™ืงืฉืชื™ ืžืืœืคื™ ืชืœืžื™ื“ื™ ืชื™ื›ื•ืŸ ืœืขื ื•ืช ืขืœื™ื”ื.
10:37
and asked thousands of high school students to answer.
163
637196
4143
10:42
Here is an example:
164
642690
1176
ื”ื ื” ื“ื•ื’ืžื”:
10:43
[Buddhism spread to ... , Christianity to ... and Oceania,
165
643890
2818
[ื”ื‘ื•ื“ื”ื™ื–ื ื”ืชืคืฉื˜ ืืœ ืืกื™ื”, ื”ื ืฆืจื•ืช ืœืื™ืจื•ืคื” ื•ืื•ืงื™ืื ื™ื”...]
10:46
and Islam to ...]
166
646732
1151
10:47
Of course, the original problems are written in Japanese,
167
647907
2740
[ืื™ื–ื• ื“ืช ื”ื’ื™ืขื” ืœืื•ืงื™ืื ื™ื”?]
ื”ื‘ืขื™ื•ืช ื ื›ืชื‘ื• ื›ืžื•ื‘ืŸ ื‘ื™ืคื ื™ืช, ืฉืคืช ื”ืื ืฉืœื”ื.
10:50
their mother tongue.
168
650671
1155
10:51
[ ______ has spread to Oceania.
169
651850
1515
10:53
1. Hinduism 2. Christianity 3. Islam 4. Buddhism ]
170
653389
2417
ื‘ืจื•ืจ ืฉื”ืชืฉื•ื‘ื” ื”ื™ื "ื ืฆืจื•ืช", ื ื›ื•ืŸ?
10:55
Obviously, Christianity is the answer, isn't it?
171
655830
2299
ื–ื” ื”ืจื™ ื›ืชื•ื‘ ื›ืืŸ!
10:58
It's written!
172
658153
1214
11:01
And Todai Robot chose the correct answer, too.
173
661482
4026
ื•ื’ื ืจื•ื‘ื•ื˜ ื˜ื•ื“ืื™ ื‘ื—ืจ ื‘ืชืฉื•ื‘ื” ื”ื ื›ื•ื ื”.
11:06
But one-third of junior high school students
174
666758
4879
ืื‘ืœ ืฉืœื™ืฉ ืžืชืœืžื™ื“ื™ ื—ื˜ื™ื‘ืช ื”ื‘ื™ื ื™ื™ื
11:11
failed to answer this question.
175
671661
2612
ื ื›ืฉืœ ื‘ืฉืืœื” ื–ืืช.
11:16
Do you think it is only the case in Japan?
176
676456
3159
ื ืจืื” ืœื›ื ืฉื–ื” ืจืง ื‘ื™ืคืŸ?
11:19
I do not think so,
177
679639
1976
ืœื“ืขืชื™, ืœื.
11:21
because Japan is always ranked among the top in OECD PISA tests,
178
681639
6371
ื›ื™ ื™ืคืŸ ืžื“ื•ืจื’ืช ืชืžื™ื“ ื’ื‘ื•ื” ื‘ืžื‘ื—ื ื™ ืคื™ื–"ื”
ืฉืœ ื”-OECD (ื”ืืจื’ื•ืŸ ืœืฉื™ืชื•ืฃ ืคืขื•ืœื” ื•ืœืคื™ืชื•ื— ื›ืœื›ืœื™),
11:28
measuring 15-year-old students' performance in mathematics,
179
688034
3927
ืฉืžื•ื“ื“ื™ื ื‘ื™ืฆื•ืขื™ื ืฉืœ ืชืœืžื™ื“ื™ื ื‘ื ื™ 15 ื‘ืžืชืžื˜ื™ืงื”, ืžื“ืขื™ื ื•ืงืจื™ืื”
11:31
science and reading
180
691985
1964
11:33
every three years.
181
693973
1636
ืžื“ื™ ืฉืœื•ืฉ ืฉื ื™ื.
11:39
We have been believing
182
699390
2053
ืชืžื™ื“ ื”ืืžื ื• ืฉื›ื•ืœื ื™ื›ื•ืœื™ื ืœืœืžื•ื“
11:41
that everybody can learn
183
701467
2043
11:43
and learn well,
184
703534
1905
ื•ืœืœืžื•ื“ ื”ื™ื˜ื‘,
11:45
as long as we provide good learning materials
185
705463
3697
ื›ืœ ืขื•ื“ ืžืกืคืงื™ื ืœื”ื ื—ื•ืžืจื™ ืœื™ืžื•ื“ ื˜ื•ื‘ื™ื
11:49
free on the web
186
709184
1455
ื—ื™ื ื, ื‘ืจืฉืช,
11:50
so that they can access through the internet.
187
710663
3069
ืฉืืœื™ื”ื ื™ื•ื›ืœื• ืœื’ืฉืช ื‘ืงืœื•ืช ื‘ืื™ื ื˜ืจื ื˜.
11:53
But such wonderful materials may benefit only those who can read well,
188
713756
5859
ืื‘ืœ ื—ื•ืžืจื™ื ืžืขื•ืœื™ื ื›ืืœื” ื™ื›ื•ืœื™ื ืœื”ื•ืขื™ืœ ืจืง ืœืžื™ ืฉื™ื•ื“ืขื™ื ืœืงืจื•ื ื”ื™ื˜ื‘,
12:00
and the percentage of those who can read well
189
720534
3935
ื•ืื—ื•ื– ื”ืงื•ืจืื™ื ื”ื™ื˜ื‘
12:04
may be much less than we expected.
190
724493
3378
ื”ื•ื ืื•ืœื™ ื ืžื•ืš ืžืฉืฆื™ืคื™ื ื•.
ื”ืื•ืคืŸ ื‘ื• ื‘ื ื™ ืื“ื ื™ื—ื™ื• ื‘ื“ื•-ืงื™ื•ื ืขื ื”ืชื‘ื•ื ื” ื”ืžืœืื›ื•ืชื™ืช
12:10
How we humans will coexist with AI
191
730040
4241
12:14
is something we have to think about carefully,
192
734305
3522
ื”ื•ื ืžืฉื”ื• ืฉื›ื“ืื™ ืฉื ื—ืฉื•ื‘ ืขืœื™ื• ื”ื™ื˜ื‘,
12:17
based on solid evidence.
193
737851
2137
ื•ืขืœ ืกืžืš ืขื“ื•ื™ื•ืช ืžื•ืฆืงื•ืช.
ื•ืขืœื™ื ื• ื’ื ืœื—ืฉื•ื‘ ืขืœ ื›ืš ื‘ืžื”ื™ืจื•ืช
12:21
At the same time, we have to think in a hurry
194
741063
3977
ื›ื™ ื”ื–ืžืŸ ื”ื•ืœืš ื•ืื•ื–ืœ.
12:25
because time is running out.
195
745064
2402
ืชื•ื“ื” ืœื›ื.
12:28
Thank you.
196
748106
1162
12:29
(Applause)
197
749626
3933
(ืžื—ื™ืื•ืช ื›ืคื™ื™ื)
12:34
Chris Anderson: Noriko, thank you.
198
754211
2080
ื›ืจื™ืก ืื ื“ืจืกื•ืŸ: ื ื•ืจื™ืงื•, ืชื•ื“ื” ืœืš.
12:36
Noriko Arai: Thank you.
199
756315
1765
ื ื•ืจื™ืงื• ืืจืื™: ืชื•ื“ื”.
ื›"ื: ื”ืžื—ืฉืช ื‘ื”ืจืฆืืชืš ื‘ืื•ืคืŸ ื ืคืœื ืื™ืš ื”ืชื‘ื•ื ื” ื”ืžืœืื›ื•ืชื™ืช ื—ื•ืฉื‘ืช,
12:38
CA: In your talk, you so beautifully give us a sense of how AIs think,
200
758104
5304
12:43
what they can do amazingly
201
763432
1564
ืžื” ื”ื™ื ื™ื›ื•ืœื” ืœืขืฉื•ืช ื ื”ื“ืจ,
ื•ืžื” ืœื.
12:45
and what they can't do.
202
765020
1695
12:46
But -- do I read you right,
203
766739
1494
ืื‘ืœ -- ื”ืื ื”ื‘ื ืชื™ ืื•ืชืš ื ื›ื•ืŸ,
12:48
that you think we really need quite an urgent revolution in education
204
768257
5270
ืฉืืช ื—ื•ืฉื‘ืช ืฉื ื—ื•ืฆื” ืœื ื• ื‘ื“ื—ื™ืคื•ืช ืžื”ืคื›ื” ื‘ื—ื™ื ื•ืš,
12:53
to help kids do the things that humans can do better than AIs?
205
773551
4155
ื›ื“ื™ ืœืขื–ื•ืจ ืœื™ืœื“ื™ื ืœื”ืฉืชืคืจ ื‘ืขืฉื™ื™ืช ื“ื‘ืจื™ื
ืฉื‘ื ื™-ืื“ื ืขื•ืฉื™ื ื˜ื•ื‘ ื™ื•ืชืจ ืžื”ืชื‘ื•ื ื” ื”ืžืœืื›ื•ืชื™ืช?
12:57
NA: Yes, yes, yes.
206
777730
1328
ื "ื: ื›ืŸ, ื›ืŸ, ื›ืŸ.
12:59
Because we humans can understand the meaning.
207
779082
4035
ื›ื™ ืื ื•, ื›ื‘ื ื™-ืื“ื, ื™ื›ื•ืœื™ื ืœื”ื‘ื™ืŸ ืžืฉืžืขื•ื™ื•ืช.
ื–ื” ื—ืกืจ ื‘ืžื™ื“ื” ืจื‘ื” ืžืื“ ื‘ืชื‘ื•ื ื” ื”ืžืœืื›ื•ืชื™ืช.
13:03
That is something which is very, very lacking in AI.
208
783141
4906
ืื‘ืœ ืจื•ื‘ ื”ืชืœืžื™ื“ื™ื ืคืฉื•ื˜ ืฆื•ื‘ืจื™ื ืืช ื”ื™ื“ืข
13:08
But most of the students just pack the knowledge
209
788071
4368
13:12
without understanding the meaning of the knowledge,
210
792463
3903
ืžื‘ืœื™ ืœื”ื‘ื™ืŸ ืืช ืžืฉืžืขื•ืช ื”ื™ื“ืข,
13:16
so that is not knowledge, that is just memorizing,
211
796390
2809
ื•ืœื›ืŸ ื–ื” ืœื ื™ื“ืข, ืืœื ืกืชื ืฉื™ื ื•ืŸ,
ืžืฉื”ื• ืฉื’ื ื”ืชื‘ื•ื ื” ื”ืžืœืื›ื•ืชื™ืช ื™ื•ื“ืขืช ืœืขืฉื•ืช.
13:19
and AI can do the same thing.
212
799223
2450
13:21
So we have to think about a new type of education.
213
801697
3631
ืœื›ืŸ ืขืœื™ื ื• ืœื—ืฉื•ื‘ ืขืœ ื—ื™ื ื•ืš ืžืกื•ื’ ืฉื•ื ื”.
ื›"ื: ืœืขื‘ื•ืจ ืžืฉื™ื ื•ืŸ ื™ื“ืข ืœื”ื‘ื ืช ืžืฉืžืขื•ืชื•.
13:25
CA: A shift from knowledge, rote knowledge, to meaning.
214
805352
3289
13:28
NA: Mm-hmm.
215
808665
1151
ื "ื: ื ื›ื•ืŸ.
13:29
CA: Well, there's a challenge for the educators. Thank you so much.
216
809840
3240
ื›"ื: ื—ืชื™ื›ืช ืืชื’ืจ ืœืื ืฉื™ ื”ื—ื™ื ื•ืš ืฉื‘ื™ื ื™ื ื•. ืชื•ื“ื” ืจื‘ื” ืœืš.
ื "ื: ืชื•ื“ื” ืจื‘ื” ืœืš. ืชื•ื“ื”.
13:33
NA: Thank you very much. Thank you.
217
813104
1698
(ืžื—ื™ืื•ืช ื›ืคื™ื™ื)
13:34
(Applause)
218
814826
1185
ืขืœ ืืชืจ ื–ื”

ืืชืจ ื–ื” ื™ืฆื™ื’ ื‘ืคื ื™ื›ื ืกืจื˜ื•ื ื™ YouTube ื”ืžื•ืขื™ืœื™ื ืœืœื™ืžื•ื“ ืื ื’ืœื™ืช. ืชื•ื›ืœื• ืœืจืื•ืช ืฉื™ืขื•ืจื™ ืื ื’ืœื™ืช ื”ืžื•ืขื‘ืจื™ื ืขืœ ื™ื“ื™ ืžื•ืจื™ื ืžื”ืฉื•ืจื” ื”ืจืืฉื•ื ื” ืžืจื—ื‘ื™ ื”ืขื•ืœื. ืœื—ืฅ ืคืขืžื™ื™ื ืขืœ ื”ื›ืชื•ื‘ื™ื•ืช ื‘ืื ื’ืœื™ืช ื”ืžื•ืฆื’ื•ืช ื‘ื›ืœ ื“ืฃ ื•ื™ื“ืื• ื›ื“ื™ ืœื”ืคืขื™ืœ ืืช ื”ืกืจื˜ื•ืŸ ืžืฉื. ื”ื›ืชื•ื‘ื™ื•ืช ื’ื•ืœืœื•ืช ื‘ืกื ื›ืจื•ืŸ ืขื ื”ืคืขืœืช ื”ื•ื•ื™ื“ืื•. ืื ื™ืฉ ืœืš ื”ืขืจื•ืช ืื• ื‘ืงืฉื•ืช, ืื ื ืฆื•ืจ ืื™ืชื ื• ืงืฉืจ ื‘ืืžืฆืขื•ืช ื˜ื•ืคืก ื™ืฆื™ืจืช ืงืฉืจ ื–ื”.

https://forms.gle/WvT1wiN1qDtmnspy7