The Mind-Reading Potential of AI | Chin-Teng Lin | TED

68,472 views ใƒป 2025-01-07

TED


์•„๋ž˜ ์˜๋ฌธ์ž๋ง‰์„ ๋”๋ธ”ํด๋ฆญํ•˜์‹œ๋ฉด ์˜์ƒ์ด ์žฌ์ƒ๋ฉ๋‹ˆ๋‹ค.

๋ฒˆ์—ญ: Alvin Choi ๊ฒ€ํ† : Ju-young Moon
00:04
How often are you frustrated by the time it takes
0
4020
5520
๋จธ๋ฆฟ์†์— ์žˆ๋Š” ๊ฒƒ๋“ค์„ ์ปดํ“จํ„ฐ๋กœ ์ •ํ™•ํ•˜๊ฒŒ ์˜ฎ๊ธฐ๋Š” ๋ฐ ๊ฑธ๋ฆฌ๋Š”
00:09
to accurately get things in your mind into a computer?
1
9580
5240
์‹œ๊ฐ„ ๋•Œ๋ฌธ์— ์–ผ๋งˆ๋‚˜ ์ž์ฃผ ๋‹ต๋‹ตํ•จ์„ ๋А๋ผ์‹œ๋‚˜์š”?
00:16
It is even worse for people like me,
2
16140
3560
์ €์ฒ˜๋Ÿผ ๊ธ€์ž๋ฅผ ๊ธฐ๋ฐ˜์œผ๋กœ ํ•˜์ง€ ์•Š๋Š” ๋ชจ๊ตญ์–ด๋ฅผ ์‚ฌ์šฉํ•˜๋Š” ์‚ฌ๋žŒ๋“ค์—๊ฒŒ๋Š”
00:19
whose first language is not based on letters.
3
19700
4200
๋” ๋‚˜์œ ์ƒํ™ฉ์ž…๋‹ˆ๋‹ค.
00:24
I live and work in Australia,
4
24740
3080
์ €๋Š” ํ˜ธ์ฃผ์— ๊ฑฐ์ฃผํ•˜๋ฉฐ ์ผํ•˜๊ณ  ์žˆ์ง€๋งŒ ์›๋ž˜๋Š” ๋Œ€๋งŒ ์ถœ์‹ ์ž…๋‹ˆ๋‹ค.
00:27
but I am originally from Taiwan.
5
27820
3280
00:32
I moved to Sydney eight years ago
6
32180
3280
์‹œ๋“œ๋‹ˆ์—๋Š” 8๋…„ ์ „์— ์ด์ฃผํ–ˆ์œผ๋ฉฐ
00:35
and now run a university research center there.
7
35500
5360
์ง€๊ธˆ์€ ์ด๊ณณ์—์„œ ๋Œ€ํ•™ ์—ฐ๊ตฌ ์„ผํ„ฐ๋ฅผ ์šด์˜ํ•˜๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
00:42
Most of us use keyboards every day
8
42300
3840
์ธ๊ฐ„์€ ๋Œ€๋ถ€๋ถ„ ๋งค์ผ ๋จธ๋ฆฟ์†์— ๋– ์˜ค๋ฅด๋Š” ๊ฒƒ๋“ค์„
00:46
to get things in our minds into the computer.
9
46180
3880
์ปดํ“จํ„ฐ๋กœ ์˜ฎ๊ธฐ๊ธฐ ์œ„ํ•ด ํ‚ค๋ณด๋“œ๋ฅผ ์‚ฌ์šฉํ•˜๊ธฐ ๋•Œ๋ฌธ์—
00:51
We have to learn to type.
10
51300
1760
ํƒ€์ดํ•‘ํ•˜๋Š” ๋ฒ•์„ ๋ฐฐ์›Œ์•ผ ํ•ฉ๋‹ˆ๋‹ค.
00:53
The fact that you have to learn to do some things
11
53740
4320
์–ด๋–ค ์ผ์„ ํ•˜๋Š” ๋ฒ•์„ ๋ฐฐ์›Œ์•ผ ํ•œ๋‹ค๋Š” ์‚ฌ์‹ค์€
00:58
shows how unnatural it is.
12
58100
3400
์–ผ๋งˆ๋‚˜ ๋ถ€์ž์—ฐ์Šค๋Ÿฌ์šด ์ผ์ธ์ง€ ๋ณด์—ฌ์ค๋‹ˆ๋‹ค.
01:02
The finger-driven touch screen has been around for 60 years.
13
62740
5640
์†๊ฐ€๋ฝ์œผ๋กœ ์ž‘๋™ํ•˜๋Š” ํ„ฐ์น˜์Šคํฌ๋ฆฐ์€ 60์—ฌ ๋…„ ๋™์•ˆ ์‚ฌ์šฉ๋˜์–ด ์™”์œผ๋ฉฐ
01:08
It's convenient, but it is also slow.
14
68780
3520
ํŽธ๋ฆฌํ•˜์ง€๋งŒ ๋А๋ฆฌ๊ธฐ๋„ ํ•˜์ฃ .
01:14
There are other ways to control computers --
15
74020
3920
์ปดํ“จํ„ฐ๋ฅผ ์ œ์–ดํ•˜๋Š” ๋‹ค๋ฅธ ๋ฐฉ๋ฒ•์œผ๋กœ๋Š”
01:17
joystick or gestures --
16
77980
2920
์กฐ์ด์Šคํ‹ฑ์ด๋‚˜ ์ œ์Šค์ฒ˜๊ฐ€ ์žˆ์ง€๋งŒ
01:20
but they are not very useful in capturing the words in your mind.
17
80940
6680
๋จธ๋ฆฟ์†์— ์žˆ๋Š” ๋‹จ์–ด๋ฅผ ํฌ์ฐฉํ•˜๋Š” ๋ฐ์—๋Š” ๊ทธ๋‹ค์ง€ ์œ ์šฉํ•˜์ง€ ์•Š์Šต๋‹ˆ๋‹ค.
01:28
And it is words --
18
88260
2680
๊ทธ๋ฆฌ๊ณ  ๋‹จ์–ด๋Š”
01:30
they are critical to communication for human beings.
19
90980
4320
์ธ๊ฐ„์˜ ์˜์‚ฌ์†Œํ†ต์— ๋งค์šฐ ์ค‘์š”ํ•˜๋ฉฐ
01:36
The problem is about to be over,
20
96980
4360
๋ฌธ์ œ๋Š” ๊ณง ํ•ด๊ฒฐ๋  ๊ฒƒ์ธ๋ฐ์š”
01:41
because of AI.
21
101380
2000
AI (์ธ๊ณต์ง€๋Šฅ) ๋•๋ถ„์ด์ฃ .
01:43
Today, I will show you
22
103980
2360
์˜ค๋Š˜ ๊ฐ•์—ฐ์„ ํ†ตํ•ด ์–ด๋–ป๊ฒŒ AI๊ฐ€
01:46
how AI can turn the speech in your mind into words on screen.
23
106340
7000
์—ฌ๋Ÿฌ๋ถ„์˜ ๋จธ๋ฆฟ์† ๋ง์„ ํ™”๋ฉด ์† ๋‹จ์–ด๋กœ ๋ฐ”๊ฟ€ ์ˆ˜ ์žˆ๋Š”์ง€ ๋ณด์—ฌ๋“œ๋ฆฌ๊ฒ ์Šต๋‹ˆ๋‹ค.
01:55
Getting from the brain to the computer efficiently
24
115980
5560
๋‘๋‡Œ์—์„œ ์ปดํ“จํ„ฐ๋กœ ํšจ์œจ์ ์œผ๋กœ ์ด๋™ํ•˜๋Š” ๊ฒƒ์€
02:01
is a real bottleneck for any computer application.
25
121580
3920
๋ชจ๋“  ์ปดํ“จํ„ฐ ์‘์šฉ ํ”„๋กœ๊ทธ๋žจ์—์„œ ์‹ค์งˆ์ ์ธ ๋ณ‘๋ชฉ ํ˜„์ƒ์œผ๋กœ์จ
02:06
It has been my passion for 25 years.
26
126460
3520
25๋…„ ๋™์•ˆ ์ œ๊ฐ€ ์Ÿ์•„๋ถ€์€ ์—ด์ •์ด์—ˆ์ฃ .
02:11
Many of you, or most of you,
27
131860
2160
์—ฌ๋Ÿฌ๋ถ„ ์ค‘ ๋งŽ์€ ๋ถ„์€ ํ˜น์€ ์—ฌ๋Ÿฌ๋ถ„ ๋Œ€๋ถ€๋ถ„์ด
02:14
have heard of "brain-computer interface," BCI.
28
134060
5040
โ€œ๋‡Œ-์ปดํ“จํ„ฐ ์ธํ„ฐํŽ˜์ด์Šคโ€์ธ BCI์— ๋Œ€ํ•ด ๋“ค์–ด๋ณด์…จ์„ ๊ฒ๋‹ˆ๋‹ค.
02:20
I have been working on BCI,
29
140100
2400
์ €๋Š” 2004๋…„๋ถ€ํ„ฐ
02:22
for the direct communication between the brain and machine,
30
142500
5080
๋‡Œ์™€ ๊ธฐ๊ณ„ ์‚ฌ์ด์˜ ์ง์ ‘์ ์ธ ์˜์‚ฌ์†Œํ†ต์„ ์œ„ํ•œ
02:27
since 2004.
31
147620
1920
BCI ์—ฐ๊ตฌ๋ฅผ ํ†ตํ•ด์„œ
02:30
I developed a series of EEG headsets that do this.
32
150460
6320
์ด๋Ÿฐ ๊ธฐ๋Šฅ์„ ํ•˜๋Š” EEG ํ—ค๋“œ์…‹ ์‹œ๋ฆฌ์ฆˆ๋ฅผ ๊ฐœ๋ฐœํ–ˆ๋Š”๋ฐ
02:38
But they are not new.
33
158780
1680
ํš๊ธฐ์ ์ธ ๊ฑด ์•„๋‹ˆ์ง€๋งŒ
02:41
What is new is an interface that works in a natural way,
34
161700
5560
๋‡Œ๊ฐ€ ์ž์—ฐ์Šค๋Ÿฝ๊ฒŒ ์ž‘๋™ํ•˜๋Š” ๋ฐฉ์‹์— ๋”ฐ๋ผ
02:47
based on how our brain is working naturally.
35
167300
4520
์ž์—ฐ์Šค๋Ÿฝ๊ฒŒ ์ž‘๋™ํ•˜๋Š” ์ธํ„ฐํŽ˜์ด์Šค๋Š” ์ƒˆ๋กœ์šด ์ ์ด์ฃ .
02:53
Imagine reading words when someone is thinking,
36
173140
5520
์ธ๊ฐ„์˜ ์ƒ๊ฐ๋งŒ์œผ๋กœ ๋‹จ์–ด๋ฅผ ์ฝ๊ณ  ๋‡Œ ์‹ ํ˜ธ๋ฅผ
02:58
translating the brain signals into words.
37
178700
3640
๋‹จ์–ด๋กœ ๋ฒˆ์—ญํ•œ๋‹ค๊ณ  ์ƒ์ƒํ•ด ๋ณด์„ธ์š”.
03:03
Today, you will see this in action,
38
183220
3320
์˜ค๋Š˜๋‚ ์—๋Š” ์ž…๋ ฅ ์ž‘์—… ์—†์ด๋„ ์‹ค์ œ๋กœ ์ด๋Ÿฐ ๋ชจ์Šต์„ ๋ณผ ์ˆ˜ ์žˆ์„ ๊ฒƒ์ž…๋‹ˆ๋‹ค.
03:06
and with no implants.
39
186580
2080
03:09
We are using AI
40
189940
1480
์šฐ๋ฆฌ๋Š” AI๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ
03:11
to decode the brain signals on the top of your head
41
191460
4200
์ธ๊ฐ„ ๋‘๋‡Œ์˜ ์‹ ํ˜ธ๋ฅผ ํ•ด๋…ํ•˜๊ณ 
03:15
and identify the biomarkers of speaking.
42
195660
3960
๋งํ•˜๋Š” ๋ฐ ํ•„์š”ํ•œ ์ƒ์ฒด์ง€ํ‘œ๋ฅผ ์ฐพ์•„๋‚ด๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
03:20
That means that you can send the words in your mind into the computer
43
200260
6200
์ฆ‰, ์›จ์–ด๋Ÿฌ๋ธ” ๊ธฐ์ˆ ์„ ํ†ตํ•ด ๋จธ๋ฆฟ์†์— ์žˆ๋Š” ๋‹จ์–ด๋ฅผ
03:26
with wearable technology.
44
206500
2400
์ปดํ“จํ„ฐ๋กœ ๋ณด๋‚ผ ์ˆ˜ ์žˆ๋‹ค๋Š” ๋œป์ด์ฃ .
03:29
It's exciting,
45
209540
1280
์ด๋Š” ์ •๋ง ํฅ๋ฏธ๋กœ์šด ์ผ์ด๋ฉฐ
03:30
and I believe it will open up the bottleneck
46
210820
4600
์šฐ๋ฆฌ๊ฐ€ ์ปดํ“จํ„ฐ๋ฅผ ์‚ฌ์šฉํ•˜๋Š” ๋ฐฉ์‹์˜ ๋ณ‘๋ชฉ ํ˜„์ƒ์„
03:35
of how we engage with computers.
47
215420
3640
์—†์•จ ์ˆ˜ ์žˆ์„ ๊ฑฐ๋กœ ์ƒ๊ฐํ•ฉ๋‹ˆ๋‹ค.
03:40
We are making exciting progress in decoding EEG to test this.
48
220940
5880
์ด๋ฅผ ์‹œํ—˜ํ•˜๋ ค๊ณ  ๋‡ŒํŒŒ ํ•ด๋…์— ๋†€๋ผ์šด ์ง„์ „์„ ์ด๋ฃจ๊ณ  ์žˆ๋Š”๋ฐ
03:47
It's natural.
49
227260
1120
๋‹น์—ฐํ•œ ๊ฒฐ๊ณผ์ฃ .
03:48
We have had very promising results
50
228980
3600
์šฐ๋ฆฌ๋Š” ๋ˆ„๊ตฐ๊ฐ€ ํฐ ์†Œ๋ฆฌ๋กœ ๋งํ•  ๋•Œ
03:52
in decoding EEG when someone is speaking aloud.
51
232620
4200
๋‡ŒํŒŒ๋ฅผ ํ•ด๋…ํ•  ์ˆ˜ ์žˆ๋‹ค๋Š” ๋งค์šฐ ์œ ๋งํ•œ ๊ฒฐ๊ณผ๋ฅผ ์–ป์—ˆ์Šต๋‹ˆ๋‹ค.
03:57
The frontier we are working on now
52
237340
2600
์šฐ๋ฆฌ๊ฐ€ ํ˜„์žฌ ์—ฐ๊ตฌํ•˜๊ณ  ์žˆ๋Š” ๋ถ„์•ผ๋Š”
03:59
is to decode EEG when the speech is not spoken aloud,
53
239980
6200
์Œ์„ฑ์ด ์—†์„ ๋•Œ๋„ ๋‡ŒํŒŒ๋ฅผ ํ•ด๋…ํ•˜๋Š” ๊ฒƒ์ธ๋ฐ
04:06
the words that flow in your mind when you are listening to others
54
246220
6120
์ด๋Š” ๋‹ค๋ฅธ ์‚ฌ๋žŒ์˜ ๋ง์„ ๋“ค์„ ๋•Œ ํ˜น์€ ์ž์‹ ์—๊ฒŒ
04:12
or when you are talking to yourself or thinking.
55
252380
4840
๋งํ•˜๊ฑฐ๋‚˜ ์ƒ๊ฐํ•  ๋•Œ ๋จธ๋ฆฟ์†์— ๋– ์˜ค๋ฅด๋Š” ๋‹จ์–ด๋“ค๋กœ์จ
04:18
We are well on the way to make it a reality.
56
258700
4240
์šฐ๋ฆฌ๋Š” ๊ทธ๊ฒƒ์„ ํ˜„์‹คํ™”ํ•˜๋Š” ๋ฐ ์ˆœ์กฐ๋กญ๊ฒŒ ์ง„ํ–‰ํ•˜๊ณ  ์žˆ์ฃ .
04:24
Who would like to see this in action?
57
264500
1960
์‹ค์ œ๋กœ ๋ณด๊ณ  ์‹ถ์€ ๋ถ„?
04:26
(Cheers and applause)
58
266780
6920
(ํ™˜ํ˜ธ์™€ ๋ฐ•์ˆ˜)
04:33
Great, we are ready to demonstrate it to you.
59
273740
3600
์ข‹์•„์š”, ์ด์ œ ๋ณด์—ฌ๋“œ๋ฆด ์ค€๋น„๊ฐ€ ๋์Šต๋‹ˆ๋‹ค.
04:37
I am going to invite two of my team, Charles and Daniel,
60
277740
4640
์ €ํฌ ํŒ€์› ์ค‘ ์ฐฐ์Šค์™€ ๋‹ค๋‹ˆ์—˜์„ ์ดˆ๋Œ€ํ•ด์„œ
04:42
to show it to us again.
61
282420
2360
๋‹ค์‹œ ๋ณด์—ฌ๋“œ๋ฆด๊ฒŒ์š”.
04:45
This is the first world premiere for us,
62
285900
3120
์ด๋ฒˆ์ด ์„ธ๊ณ„ ์ตœ์ดˆ๋กœ ๊ณต๊ฐœ๋˜๋Š” ๊ฒƒ์ด๋‹ˆ
04:49
so I hope you can be patient with us.
63
289060
3960
์ธ๋‚ด์‹ฌ์„ ๊ฐ–๊ณ  ๊ธฐ๋‹ค๋ ค ์ฃผ์…จ์œผ๋ฉด ์ข‹๊ฒ ์Šต๋‹ˆ๋‹ค.
04:53
We are getting around 50 percent accuracy ...
64
293060
3800
์ •ํ™•๋„๋Š” ์•ฝ 50% ์ •๋„์ด์ง€๋งŒ...
04:56
(Laughter)
65
296900
2280
(์›ƒ์Œ)
04:59
in decoding the brain signals into words
66
299220
3960
๋ˆ„๊ตฐ๊ฐ€ ์กฐ์šฉํžˆ ๋งํ•  ๋•Œ
05:03
when someone is speaking silently.
67
303220
3680
๋‡Œ ์‹ ํ˜ธ๋ฅผ ๋ง๋กœ ํ•ด๋…ํ•˜๋Š” ๊ฑฐ์ฃ .
05:07
Here shows how it will work.
68
307860
2080
์–ด๋–ป๊ฒŒ ์ž‘๋™ํ•˜๋Š”์ง€ ๋ณด์—ฌ๋“œ๋ฆด๊ฒŒ์š”.
05:12
We have a collection of words that we have trained our technology with.
69
312860
5760
์ €ํฌ๋Š” ๊ธฐ์ˆ ์„ ํ•™์Šต์‹œํ‚ค๋Š” ๋ฐ ์‚ฌ์šฉํ•œ ๋‹จ์–ด ๋ชจ์Œ์„ ๊ฐ€์ง€๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
05:20
They are combined into sentences.
70
320260
2640
์ด ๋‘ ๋ฌธ์žฅ์ด ํ•ฉ์ณ์ ธ ๋ฌธ์žฅ์œผ๋กœ ๋งŒ๋“ค์–ด์ง‘๋‹ˆ๋‹ค.
05:24
Charles will select one sentence,
71
324180
2760
์ฐฐ์Šค๊ฐ€ ํ•œ ๋ฌธ์žฅ์„ ์„ ํƒํ•˜๋ฉด
05:26
and Daniel will read the sentence word by word, silently,
72
326940
5240
๋‹ค๋‹ˆ์—˜์€ ์กฐ์šฉํžˆ ๊ทธ ๋ฌธ์žฅ์„ ํ•œ ๋‹จ์–ด์”ฉ ์ฝ๊ณ 
05:32
and produce the brain signals that will be picked up by our sensors.
73
332180
5280
์„ผ์„œ๊ฐ€ ๊ฐ์ง€ํ•˜๋Š” ๋‡Œ ์‹ ํ˜ธ๋ฅผ ์ƒ์„ฑํ•ฉ๋‹ˆ๋‹ค.
05:37
Our technology will decode the brain signals into words.
74
337460
4760
์ด ๊ธฐ์ˆ ์€ ๋‡Œ ์‹ ํ˜ธ๋ฅผ ๋‹จ์–ด๋กœ ํ•ด๋…ํ•  ๊ฒƒ์ž…๋‹ˆ๋‹ค.
05:42
Charles, Daniel, are you ready to go ahead?
75
342940
2760
์ฐฐ์Šค, ๋‹ค๋‹ˆ์—˜ ๊ณ„์†ํ•  ์ค€๋น„๊ฐ€ ๋˜์…จ๋‚˜์š”?
05:48
This is the sentence
76
348460
3800
๋‹ค๋‹ˆ์—˜์ด ์กฐ์šฉํžˆ ์ฝ์„ ๋ฌธ์žฅ์€
05:52
that Daniel is going to read silently.
77
352300
3560
์ด ๋‹จ์–ด์ž…๋‹ˆ๋‹ค.
06:15
(Applause)
78
375860
5360
(๋ฐ•์ˆ˜)
06:21
Sorry, please keep silent. (Laughter)
79
381980
2800
์ฃ„์†กํ•ฉ๋‹ˆ๋‹ค. ์กฐ์šฉํžˆ ํ•ด์ฃผ์„ธ์š”. (์›ƒ์Œ)
06:24
Here are the --
80
384820
1760
์—ฌ๊ธฐ์—...
06:27
decoded words.
81
387940
1520
ํ•ด๋…๋œ ๋‹จ์–ด๋“ค์ด ์žˆ์–ด์š”.
06:29
They are likely the intended words.
82
389500
2320
์˜๋„๋œ ๋‹จ์–ด์ผ ๊ฐ€๋Šฅ์„ฑ์ด ๋†’์Šต๋‹ˆ๋‹ค.
06:32
You can see the probability ranking
83
392540
2520
๋‹น์‚ฌ์˜ ๊ธฐ์ˆ ์„ ํ†ตํ•ด
06:35
of the decoded words by our technology.
84
395100
3080
๋””์ฝ”๋”ฉ๋œ ๋‹จ์–ด์˜ ํ™•๋ฅ  ์ˆœ์œ„๋ฅผ ํ™•์ธํ•  ์ˆ˜ ์žˆ๋Š”๋ฐ์š”.
06:40
The pattern shows our predicted sentence ...
85
400540
3880
์ด ํŒจํ„ด์€ ์ €ํฌ ์˜ˆ์ธก ๋ฌธ์žฅ์„ ๋ณด์—ฌ์ฃผ๋Š”๋ฐ...
06:45
is not so correct.
86
405820
1680
์ •ํ™•ํ•˜์ง„ ์•Š์ฃ .
06:47
(Laughter)
87
407540
3080
(์›ƒ์Œ)
06:50
Sorry, you see the other 50 percent,
88
410620
4800
์ฃ„์†กํ•ฉ๋‹ˆ๋‹ค. ๋‚˜๋จธ์ง€ 50% ๋Š” ๋ณด์ด๋Š”๋ฐ
06:55
the system doesn't work.
89
415420
1240
์‹œ์Šคํ…œ์ด ์ž‘๋™ํ•˜์ง€ ์•Š์•„์š”.
06:56
But actually, you still can see some keywords where we got it.
90
416660
3800
ํ•˜์ง€๋งŒ ์‹ค์ œ๋กœ ์–ด๋””์„œ ๊ฐ€์ ธ์˜จ ํ‚ค์›Œ๋“œ๋„ ๋ช‡ ๊ฐœ ๋ณด์ผ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
07:01
Let's invite Charles and Daniel to do it again,
91
421180
3480
์ฐฐ์Šค์™€ ๋‹ค๋‹ˆ์—˜์„ ์ดˆ๋Œ€ํ•ด ๋‹ค์‹œ ํ•œ๋ฒˆ ํ•ด๋ณด์ฃ .
07:04
but please keep silent when he's reading silently.
92
424700
4640
ํ•˜์ง€๋งŒ ๊ทธ๊ฐ€ ์กฐ์šฉํžˆ ๊ธ€์„ ์ฝ์„ ๋•Œ๋Š” ์กฐ์šฉํžˆ ์žˆ์–ด ์ฃผ์„ธ์š”.
07:10
Here is the sentence,
93
430660
1320
์ด ๋ฌธ์žฅ์€ ๋‹ค๋‹ˆ์—˜์ด ์กฐ์šฉํžˆ ํ•œ ๊ธ€์ž์”ฉ ์ฝ์„ ๋˜ ๋‹ค๋ฅธ ๋ฌธ์žฅ์ž…๋‹ˆ๋‹ค.
07:12
another sentence that Daniel will read word by word, silently.
94
432020
4240
07:16
(Laughs)
95
436980
1160
(์›ƒ์Œ)
07:45
Again, here are the decoded words.
96
465620
2480
๋‹ค์‹œ ๋ง์”€๋“œ๋ฆฌ์ง€๋งŒ ํ•ด๋…๋œ ๋‹จ์–ด๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™์Šต๋‹ˆ๋‹ค.
07:48
They are likely the intended words.
97
468100
2520
์˜๋„ํ•œ ๋‹จ์–ด์ผ ๊ฐ€๋Šฅ์„ฑ์ด ๋†’์ฃ .
07:53
The pattern shows our predictive sentence
98
473260
3640
์ด ํŒจํ„ด์€ ์ €ํฌ๊ฐ€ ์˜ˆ์ธกํ•œ ๋ฌธ์žฅ์ด
07:56
is very close to the ground-truth sentence this time.
99
476940
3360
์ด๋ฒˆ์—๋Š” ์‹ค์กด ๋ฌธ์žฅ๊ณผ ๋งค์šฐ ๋น„์Šทํ•˜๋‹ค๋Š” ๊ฒƒ์„ ๋ณด์—ฌ์ค๋‹ˆ๋‹ค.
08:00
(Cheers and applause)
100
480300
6920
(ํ™˜ํ˜ธ์™€ ๋ฐ•์ˆ˜)
08:07
Thank you, thank you.
101
487620
1840
๊ฐ์‚ฌํ•ฉ๋‹ˆ๋‹ค, ๊ฐ์‚ฌํ•ฉ๋‹ˆ๋‹ค.
08:10
How does it work?
102
490340
1560
์–ด๋–ป๊ฒŒ ์ž‘๋™ํ•˜๋‚˜์š”?
08:11
We pick up the brain signals with sensors
103
491940
4200
์„ผ์„œ๋กœ ๋‡Œ ์‹ ํ˜ธ๋ฅผ
08:16
and amplify and filter them
104
496140
3560
ํฌ์ฐฉํ•˜๊ณ , ์ฆํญํ•˜๊ณ , ํ•„ํ„ฐ๋งํ•˜์—ฌ
08:19
to reduce the noise and get the right biomarkers.
105
499700
4040
์†Œ์Œ์„ ์ค„์ด๊ณ  ์ ์ ˆํ•œ ๋ฐ”์ด์˜ค๋งˆ์ปค๋ฅผ ์–ป๋Š”๋ฐ
08:24
We use AI for the task.
106
504460
2880
๋ฐ”๋กœ ์ด ์ž‘์—…์— AI๋ฅผ ์‚ฌ์šฉํ•˜์ฃ .
08:27
We use deep learning to decode the brain signals
107
507380
4080
์šฐ๋ฆฌ๋Š” ๋”ฅ๋Ÿฌ๋‹์„ ์‚ฌ์šฉํ•˜์—ฌ ๋‡Œ ์‹ ํ˜ธ๋ฅผ ์˜๋„ํ•œ ๋‹จ์–ด๋กœ ํ•ด๋…ํ•œ ํ›„
08:31
into the intended words.
108
511460
2240
08:33
And then we use the large language model
109
513740
3600
๋Œ€ํ˜• ์–ธ์–ด ๋ชจ๋ธ์„ ์‚ฌ์šฉํ•˜์—ฌ ํ•ด๋…๋œ ๋‹จ์–ด๋ฅผ ์ผ์น˜์‹œํ‚ค๊ณ 
08:37
to make the match of the decoded words
110
517380
3280
08:40
and make up for the mistakes in EEG decoding.
111
520700
4320
๋‡ŒํŒŒ ํ•ด๋…์˜ ์‹ค์ˆ˜๋ฅผ ๋ณด์™„ํ•˜๋Š”๋ฐ
08:46
All of this is going on in the AI,
112
526380
3520
์ด ๋ชจ๋“  ๊ณผ์ •์€ AI์—์„œ ์ผ์–ด๋‚˜๊ณ  ์žˆ์ง€๋งŒ
08:49
but for the user, the interaction is natural,
113
529940
3720
์‚ฌ์šฉ์ž์™€๋Š” ์ƒ๊ฐ๊ณผ ์ž์—ฐ์–ด๋ฅผ ํ†ตํ•œ
08:53
through thoughts and natural language.
114
533700
3480
์ƒํ˜ธ ์ž‘์šฉ์ด ์ž์—ฐ์Šค๋Ÿฝ์Šต๋‹ˆ๋‹ค.
08:57
We are very excited about the advances that we are making
115
537900
4960
๋‹จ์–ด์™€ ๋ฌธ์žฅ์„ ์ดํ•ดํ•˜๋Š” ๋ฐ ์žˆ์–ด
09:02
in understanding words and sentences.
116
542860
2680
์šฐ๋ฆฌ๊ฐ€ ์ด๋ฃฉํ•˜๊ณ  ์žˆ๋Š” ๋ฐœ์ „์ด ์ •๋ง ๊ธฐ๋Œ€๋ฉ๋‹ˆ๋‹ค.
09:07
Another thing that is very natural to people
117
547420
4320
์‚ฌ๋žŒ๋“ค์ด ๋งค์šฐ ์ž์—ฐ์Šค๋Ÿฝ๊ฒŒ ๋А๋ผ๋Š” ๋˜ ๋‹ค๋ฅธ ์ผ์€
09:11
is looking at something that has their attention.
118
551780
4800
๋ฐ”๋กœ ๊ด€์‹ฌ์„ ๋„๋Š” ๋Œ€์ƒ์„ ๋ฐ”๋ผ๋ณด๋Š” ๊ฒƒ์ž…๋‹ˆ๋‹ค.
09:16
Imagine if you could select an item just by looking at it,
119
556580
6520
์„ ๋ฐ˜์—์„œ ๋ฌผ๊ฑด์„ ๊ณ ๋ฅด๊ฑฐ๋‚˜
09:23
not by picking it off the shelf
120
563140
2800
์žํŒ๊ธฐ์— ์ฝ”๋“œ๋ฅผ ๋„ฃ๋Š” ๊ฒƒ์ด ์•„๋‹ˆ๋ผ
09:25
or punching a code into the vending machine.
121
565940
3760
๋ณด๊ธฐ๋งŒ ํ•ด์„œ ๋ฌผ๊ฑด์„ ์„ ํƒํ•  ์ˆ˜ ์žˆ๋‹ค๊ณ  ์ƒ์ƒํ•ด ๋ณด์„ธ์š”.
09:30
Two years ago,
122
570460
1440
2๋…„ ์ „
09:31
in a project about hands-free control of robots,
123
571900
6120
๋กœ๋ด‡์˜ ํ•ธ์ฆˆํ”„๋ฆฌ ์ œ์–ด์— ๊ด€ํ•œ ํ”„๋กœ์ ํŠธ์—์„œ
09:38
we were very excited about robot control
124
578060
4360
์ €ํฌ๋Š” ์†๊ฐ€๋ฝ์˜ ์‹œ๊ฐ์  ์‹๋ณ„์„ ํ†ตํ•œ
09:42
via visual identification of the fingers.
125
582460
4120
๋กœ๋ด‡ ์ œ์–ด์— ๋Œ€ํ•ด ๋งค์šฐ ํฅ๋ฏธ๋ฅผ ๋А๊ผˆ์Šต๋‹ˆ๋‹ค.
09:47
We are now beyond that.
126
587060
2440
์ด์ œ ๊ทธ ์ด์ƒ์„ ๋„˜์–ด์„ฐ์„ ๋ฟ ์•„๋‹ˆ๋ผ ์†๊ฐ€๋ฝ ๋™์ž‘๋„ ํ•„์š” ์—†์–ด์กŒ์ฃ .
09:49
We need not any fingers.
127
589540
2280
09:51
The AI is making it natural.
128
591860
2920
๋Œ€์‹ ์— AI๊ฐ€ ์ถ”๊ฐ€๋์œผ๋‹ˆ๊นŒ์š”.
09:55
There are four objects on the table.
129
595660
2640
์ด ํ…Œ์ด๋ธ” ์œ„์—๋Š” ๋„ค ๊ฐœ์˜ ๋ฌผ์ฒด๊ฐ€ ์žˆ์Šต๋‹ˆ๋‹ค.
09:59
Toy car,
130
599300
1960
์žฅ๋‚œ๊ฐ ์ž๋™์ฐจ์™€ ์žฅ๋‚œ๊ฐ ๋™๋ฌผ ๊ทธ๋ฆฌ๊ณ 
10:01
toy animal,
131
601260
2520
10:03
plastic flower and some food,
132
603820
3560
ํ”Œ๋ผ์Šคํ‹ฑ ๊ฝƒ ๊ทธ๋ฆฌ๊ณ  ์˜ค๋Š˜ ์•„์นจ ์‹์‚ฌ์—์„œ
10:07
which is also plastic,
133
607380
2880
10:10
not left over from the breakfast this morning.
134
610300
3480
์•ˆ ๋จน์€ ํ”Œ๋ผ์Šคํ‹ฑ ์Œ์‹๊นŒ์ง€
10:14
You can also see the four objects' photo on the screen.
135
614660
5400
ํ™”๋ฉด์—์„œ ๋„ค ๊ฐœ์˜ ๋ฌผ์ฒด ์‚ฌ์ง„ ์—ญ์‹œ ๋ณผ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
10:20
Daniel is going to look at the photos,
136
620980
3680
๋‹ค๋‹ˆ์—˜์€ ์‚ฌ์ง„์„ ๋ณด๊ณ  ๋จธ๋ฆฟ์†์— ๋– ์˜ค๋ฅด๋Š”
10:24
and select an item in his mind.
137
624700
3560
์•„์ดํ…œ์„ ํ•˜๋‚˜ ๊ณ ๋ฅด๋ ค๊ณ  ํ•ฉ๋‹ˆ๋‹ค.
10:28
If it is working as it should,
138
628860
2920
๋งŒ์ผ ์ •์ƒ์ ์œผ๋กœ ์ž‘๋™ํ•˜๋ฉด
10:31
you will see the selected item pop up on screen.
139
631820
4600
์„ ํƒํ•œ ํ•ญ๋ชฉ์ด ํ™”๋ฉด์— ๋‚˜ํƒ€๋‚˜๋Š” ๊ฒƒ์„ ๋ณผ ์ˆ˜ ์žˆ์„ ๊ฑฐ์˜ˆ์š”.
10:37
We use photos for this because they are very controllable.
140
637540
3240
์ œ์–ด๊ฐ€ ์šฉ์ดํ•˜๊ธฐ ๋•Œ๋ฌธ์— ์ €ํฌ๋Š” ์‚ฌ์ง„์„ ์‚ฌ์šฉํ•ฉ๋‹ˆ๋‹ค.
10:41
To show that this is not all just built into my presentation,
141
641900
6320
์ด ๋ชจ๋“  ๊ฒƒ์ด ์ œ ๋ฐœํ‘œ์—๋งŒ ํฌํ•จ๋˜์ง€ ์•Š๋‹ค๋Š” ๊ฒƒ์„ ๋ณด์—ฌ์ฃผ๊ธฐ ์œ„ํ•ด
10:48
Charles will pick up one item for Daniel to select in mind.
142
648220
6920
์ฐฐ์Šค๋Š” ๋‹ค๋‹ˆ์—˜์ด ์—ผ๋‘์— ๋‘์–ด์•ผ ํ•  ํ•ญ๋ชฉ ํ•˜๋‚˜๋ฅผ ๊ณจ๋ผ์ค„ ๊ฒƒ์ž…๋‹ˆ๋‹ค.
10:55
Please, Charles.
143
655460
1240
์ฐฐ์Šค!
11:01
It's a car.
144
661900
2720
์ด๊ฑด ์ฐจ์˜ˆ์š”.
11:06
So, Daniel, select ...
145
666140
2160
๋‹ค๋‹ˆ์—˜, ์„ ํƒํ•ด์š”...
11:11
the car in his mind.
146
671060
1440
๋จธ๋ฆฟ์†์— ์žˆ๋Š” ์ž๋™์ฐจ...
11:18
(Laughter)
147
678260
2360
(์›ƒ์Œ)
11:20
Hamburger.
148
680660
1200
ํ–„๋ฒ„๊ฑฐ.
11:21
(Laughter)
149
681860
1000
(์›ƒ์Œ)
11:22
It's incorrect.
150
682900
1240
ํ‹€๋ ธ์–ด์š”.
11:24
(Laughter)
151
684140
1960
(์›ƒ์Œ)
11:26
It's unlucky that the 30-percent error rate came with us again.
152
686500
5960
๋‹ค์‹œ 30%์˜ ์˜ค๋ฅ˜์œจ์ด ๋‚˜์™”๋‹ค๋Š” ๊ฑด ๋ถˆ์šดํ•œ ์ผ์ด์ฃ .
11:32
Let's invite Charles and Daniel to show it again.
153
692780
2760
์ฐฐ์Šค์™€ ๋‹ค๋‹ˆ์—˜์„ ์ดˆ๋Œ€ํ•ด์„œ ๋‹ค์‹œ ํ•œ๋ฒˆ ๋ณด์—ฌ๋“œ๋ฆด๊ฒŒ์š”.
11:40
It's a duck, a lovely duck.
154
700420
3720
์˜ค๋ฆฌ์˜ˆ์š”, ์‚ฌ๋ž‘์Šค๋Ÿฝ์ฃ .
11:44
(Laughter)
155
704140
2280
(์›ƒ์Œ)
11:51
OK. Good.
156
711780
1320
์ž, ์ข‹์•„์š”.
11:53
(Cheers and applause)
157
713100
6400
(ํ™˜ํ˜ธ์™€ ๋ฐ•์ˆ˜)
11:59
Thank you. Thank you.
158
719500
1400
๊ฐ์‚ฌํ•ฉ๋‹ˆ๋‹ค. ๊ณ ๋งˆ์›Œ์š”.
12:01
Daniel did this for his PhD.
159
721580
3000
๋‹ค๋‹ˆ์—˜์€ ๋ฐ•์‚ฌ ํ•™์œ„๋ฅผ ์œ„ํ•ด ์ด ์ผ์„ ํ–ˆ์–ด์š”.
12:04
It's very impressive, isn't it?
160
724580
1720
์ •๋ง ์ธ์ƒ์ ์ด์ง€ ์•Š๋‚˜์š”?
12:06
When Daniel selects an item in his mind,
161
726660
3480
๋‹ค๋‹ˆ์—˜์ด ๋จธ๋ฆฟ์†์—์„œ ์–ด๋–ค ํ•ญ๋ชฉ์„ ์„ ํƒํ•˜๋ฉด
12:10
his brain recognizes and identifies the object and triggers his EEGs.
162
730180
6640
๋‡Œ๊ฐ€ ๊ทธ ๋ฌผ์ฒด๋ฅผ ์ธ์‹ํ•˜๊ณ  ์‹๋ณ„ํ•˜์—ฌ ๋‡ŒํŒŒ๋ฅผ ์ž‘๋™์‹œํ‚ต๋‹ˆ๋‹ค.
12:16
Our technology decodes the triggers.
163
736860
3240
์ €ํฌ ๊ธฐ์ˆ ์€ ํŠธ๋ฆฌ๊ฑฐ๋ฅผ ํ•ด๋…ํ•ฉ๋‹ˆ๋‹ค.
12:21
We are working on our way through the technical challenges.
164
741300
6320
์ €ํฌ๋Š” ๊ธฐ์ˆ ์  ๋‚œ์ œ๋ฅผ ๊ทน๋ณตํ•˜๊ธฐ ์œ„ํ•ด ๋…ธ๋ ฅํ•˜๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
12:28
We will work on overcoming the interference issue.
165
748660
3960
๊ทธ๋ฆฌ๊ณ , ์ €ํฌ๋Š” ๊ฐ„์„ญ ๋ฌธ์ œ๋ฅผ ๊ทน๋ณตํ•˜๊ธฐ ์œ„ํ•ด ๋…ธ๋ ฅํ•  ๊ฒƒ์ž…๋‹ˆ๋‹ค.
12:33
That's why I asked for the phones to be turned off.
166
753020
4040
๊ทธ๋ž˜์„œ ์ „ํ™”๊ธฐ๋ฅผ ๊บผ๋‹ฌ๋ผ๊ณ  ์š”์ฒญํ•œ ๊ฑฐ์˜ˆ์š”.
12:38
Different people have different neural signatures,
167
758900
4560
์‚ฌ๋žŒ๋งˆ๋‹ค ์‹ ๊ฒฝ ์‹ ํ˜ธ๊ฐ€ ๋‹ค๋ฅด๊ธฐ ๋•Œ๋ฌธ์—
๋””์ฝ”๋”ฉ ์ •ํ™•๋„์— ๋งค์šฐ ์ค‘์š”ํ•ฉ๋‹ˆ๋‹ค.
12:43
which are important to decoding accuracy.
168
763500
3440
12:46
One reason I brought Daniel along here
169
766980
3640
์ œ๊ฐ€ ๋‹ค๋‹ˆ์—˜์„ ์—ฌ๊ธฐ์— ๋ฐ๋ ค์˜จ ํ•œ ๊ฐ€์ง€ ์ด์œ ๋Š”
12:50
is because he can give off great neural signatures.
170
770660
5000
๊ทธ๊ฐ€ ํ›Œ๋ฅญํ•œ ์‹ ๊ฒฝ ์‹ ํ˜ธ๋ฅผ ๋‚ผ ์ˆ˜ ์žˆ๊ธฐ ๋•Œ๋ฌธ์ž…๋‹ˆ๋‹ค.
12:55
(Laughter)
171
775700
3240
(์›ƒ์Œ)
12:58
(Applause)
172
778980
6160
(๋ฐ•์ˆ˜)
13:05
Yeah, he can give us a great neural signature,
173
785180
2760
๋„ค, ๊ทธ๋Š” ์šฐ๋ฆฌ ๊ธฐ์ˆ ๊ณผ ๊ด€๋ จ๋˜๊ธฐ๋งŒ ํ•˜๋ฉด ํ›Œ๋ฅญํ•œ ์‹ ๊ฒฝ ์‹ ํ˜ธ๋ฅผ ์ค„ ์ˆ˜ ์žˆ์–ด์š”.
13:07
as far as our technology is concerned.
174
787980
2600
13:11
They are still cables here as well.
175
791740
3120
์—ฌ๊ธฐ ์ผ€์ด๋ธ”์ด ์—ฌ์ „ํžˆ ์žˆ๊ธฐ๋„ ํ•˜์ฃ .
13:14
It is not yet very portable.
176
794900
2600
์•„์ง ํœด๋Œ€์„ฑ์ด ๊ทธ๋‹ค์ง€ ์ข‹์ง€๋Š” ์•Š์Šต๋‹ˆ๋‹ค.
13:17
Probably one biggest barrier to people using this
177
797860
6200
์ด ์ œํ’ˆ์„ ์‚ฌ์šฉํ•˜๋Š” ์‚ฌ๋žŒ๋“ค์—๊ฒŒ ํฐ ์žฅ๋ฒฝ ์ค‘ ํ•˜๋‚˜๋Š”
13:24
will be: โ€œHow do I turn it off?โ€
178
804100
2960
์•„๋งˆ๋„ โ€œ์–ด๋–ป๊ฒŒ ์ „์›์„ ๋„๋‚˜์š”?โ€ ์ผ ๊ฒ๋‹ˆ๋‹ค.
13:27
Any one of you will have had times when you are happy
179
807620
5600
์—ฌ๋Ÿฌ๋ถ„ ์ค‘ ๋ˆ„๊ตฌ๋ผ๋„ ํ•จ๊ป˜ ์žˆ๋Š” ์‚ฌ๋žŒ๋“ค์ด
13:33
that the people you are with don't know what you are really thinking.
180
813260
4280
์ž์‹ ์ด ์ง„์ • ๋ฌด์Šจ ์ƒ๊ฐ์„ ํ•˜๋Š”์ง€ ๋ชจ๋ฅธ๋‹ค๋ฉด ๊ธฐ๋ปํ•  ๊ฒ๋‹ˆ๋‹ค.
13:37
(Laughter)
181
817580
1280
(์›ƒ์Œ)
13:38
There are serious privacy and ethical issues
182
818900
2800
์šฐ๋ฆฌ ์ฃผ๋ณ€์—๋Š” ๋‹ค๋ฃจ์–ด์•ผ ํ•  ์‹ฌ๊ฐํ•œ ์‚ฌ์ƒํ™œ๊ณผ
13:41
that will have to be dealt with.
183
821740
2120
์œค๋ฆฌ์  ๋ฌธ์ œ๊ฐ€ ์žˆ์Šต๋‹ˆ๋‹ค.
13:44
I am very passionate about how important this technology can be.
184
824700
6520
์ €๋Š” ์ด ๊ธฐ์ˆ ์ด ์–ผ๋งˆ๋‚˜ ์ค‘์š”ํ•œ์ง€์— ๋Œ€ํ•ด ๋งค์šฐ ์—ด์ •์ ์ž…๋‹ˆ๋‹ค.
13:51
One exciting point is linking the brain-computer interface
185
831580
4960
ํฅ๋ฏธ๋กœ์šด ์  ์ค‘ ํ•˜๋‚˜๋Š”, ๋‡Œ-์ปดํ“จํ„ฐ ์ธํ„ฐํŽ˜์ด์Šค๋ฅผ
13:56
to wearable computers.
186
836580
2120
์›จ์–ด๋Ÿฌ๋ธ” ์ปดํ“จํ„ฐ์— ์—ฐ๊ฒฐํ•œ๋‹ค๋Š” ๊ฒƒ์ž…๋‹ˆ๋‹ค.
13:58
You already have a computer on your head.
187
838700
2720
์—ฌ๋Ÿฌ๋ถ„์˜ ๋จธ๋ฆฌ ์œ„์—๋Š” ์ด๋ฏธ ์ปดํ“จํ„ฐ๊ฐ€ ์žˆ์Šต๋‹ˆ๋‹ค.
14:01
The brain will be a natural interface.
188
841780
3240
๋‡Œ๋Š” ์ž์—ฐ์Šค๋Ÿฌ์šด ์ธํ„ฐํŽ˜์ด์Šค๊ฐ€ ๋  ๊ฒƒ์ž…๋‹ˆ๋‹ค.
14:05
It is not only about controlling a computer.
189
845060
3800
์ปดํ“จํ„ฐ ์ œ์–ด์—๋งŒ ๊ตญํ•œ๋˜์ง€ ์•Š์Šต๋‹ˆ๋‹ค.
14:09
The natural BCI also provides
190
849780
3920
๋˜ํ•œ ์ž์—ฐ์Šค๋Ÿฌ์šด BCI๋Š” ์‚ฌ๋žŒ๋“ค์ด ์„œ๋กœ ์†Œํ†ตํ•  ์ˆ˜ ์žˆ๋Š”
14:13
another way for people to communicate with people.
191
853700
4040
๋˜ ๋‹ค๋ฅธ ๋ฐฉ๋ฒ•์„ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค.
14:18
For example, it allows people who are not able to speak
192
858700
5400
์˜ˆ๋ฅผ ๋“ค์–ด, ๋ง์„ ํ•  ์ˆ˜ ์—†๋Š” ์‚ฌ๋žŒ์ด๋ผ๋„
14:24
to communicate with others,
193
864100
2680
๋‹ค๋ฅธ ์‚ฌ๋žŒ๊ณผ ์˜์‚ฌ์†Œํ†ต์„ ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
14:26
or such as when privacy or silence are required.
194
866780
6520
ํ”„๋ผ์ด๋ฒ„์‹œ๋‚˜ ์นจ๋ฌต์ด ํ•„์š”ํ•  ๋•Œ ๋ง์ด์ฃ .
14:34
If your idea of nature is a lovely forest,
195
874420
6240
์—ฌ๋Ÿฌ๋ถ„์ด ์ž์—ฐ์„ ์•„๋ฆ„๋‹ค์šด ์ˆฒ์ด๋ผ๊ณ  ์ƒ๊ฐํ•œ๋‹ค๋ฉด
14:40
you could wonder how natural this could be.
196
880660
4680
์ด ์ˆฒ์ด ์–ผ๋งˆ๋‚˜ ์ž์—ฐ์Šค๋Ÿฌ์šธ์ง€ ๊ถ๊ธˆํ•  ๊ฒƒ์ž…๋‹ˆ๋‹ค.
14:45
My answer is, it's natural language,
197
885980
4560
์ œ ๋Œ€๋‹ต์€ ์ž์—ฐ์ ์ธ ๋ง์ด๊ณ 
14:50
it's the natural thought process that you are using.
198
890580
4200
์—ฌ๋Ÿฌ๋ถ„์ด ์‚ฌ์šฉํ•˜๊ณ  ์žˆ๋Š” ์ž์—ฐ์Šค๋Ÿฌ์šด ์‚ฌ๊ณ  ๊ณผ์ •์ด๋ผ๋Š” ๊ฒƒ์ž…๋‹ˆ๋‹ค.
14:54
There are no unnatural implants in your body.
199
894820
4720
์—ฌ๋Ÿฌ๋ถ„์˜ ๋ชธ์—๋Š” ๋ถ€์ž์—ฐ์Šค๋Ÿฌ์šด ์ž„ํ”Œ๋ž€ํŠธ๊ฐ€ ์—†์Šต๋‹ˆ๋‹ค.
15:00
I am challenging you to think
200
900660
2200
์ €๋Š” ์—ฌ๋Ÿฌ๋ถ„์ด ์ƒ๊ฐํ•˜๋Š” ์ž์—ฐ์Šค๋Ÿฌ์šด ์˜์‚ฌ์†Œํ†ต์— ๋Œ€ํ•ด
15:02
about what you regard as natural communication.
201
902900
5240
์ƒ๊ฐํ•ด ๋ณด๋ผ๊ณ  ๊ถŒํ•˜๊ณ  ์‹ถ์Šต๋‹ˆ๋‹ค.
15:09
Turning the speech in your mind into words.
202
909100
5640
๋จธ๋ฆฟ์†์— ์žˆ๋Š” ๋ง์„ ์‹ค์ œ์˜ ๋ง๋กœ ๋ฐ”๊พธ์„ธ์š”.
15:14
There is a standard way to finish up when talking with people --
203
914780
5240
์‚ฌ๋žŒ๋“ค๊ณผ ๋Œ€ํ™”๋ฅผ ๋๋งบ๋Š” ๋ณดํŽธ์ ์ธ ๋ฐฉ๋ฒ•์œผ๋กœ
15:20
you say: โ€œJust think about it.โ€
204
920060
4040
โ€œ๊ทธ๋ƒฅ ์ƒ๊ฐํ•ด ๋ณด์„ธ์š”.โ€๋ผ๊ณ  ํ•˜์„ธ์š”.
15:24
I hope you are as excited as we are
205
924940
3960
์—ฌ๋Ÿฌ๋ถ„์ด ๋ฌด์–ธ๊ฐ€๋ฅผ ์ƒ๊ฐํ•˜๊ธฐ๋งŒ ํ•˜๋ฉด
15:28
for the prospect of a future
206
928940
3400
๋จธ๋ฆฟ์†์— ์žˆ๋Š” ๋‹จ์–ด๋“ค์ด
15:32
in which, when you just think about something,
207
932380
4200
ํ™”๋ฉด์— ๋– ์˜ค๋ฅด๋Š” ๋ฏธ๋ž˜์— ๋Œ€ํ•œ ์ „๋ง์ด
15:36
the words in your mind appear on screen.
208
936620
4480
์ €ํฌ๋งŒํผ์ด๋‚˜ ๊ธฐ๋Œ€๊ฐ€ ๋˜์…จ์œผ๋ฉด ์ข‹๊ฒ ์Šต๋‹ˆ๋‹ค.
15:41
Thank you.
209
941140
1160
๊ฐ์‚ฌํ•ฉ๋‹ˆ๋‹ค.
15:42
(Applause)
210
942340
1760
(๋ฐ•์ˆ˜)
์ด ์›น์‚ฌ์ดํŠธ ์ •๋ณด

์ด ์‚ฌ์ดํŠธ๋Š” ์˜์–ด ํ•™์Šต์— ์œ ์šฉํ•œ YouTube ๋™์˜์ƒ์„ ์†Œ๊ฐœํ•ฉ๋‹ˆ๋‹ค. ์ „ ์„ธ๊ณ„ ์ตœ๊ณ ์˜ ์„ ์ƒ๋‹˜๋“ค์ด ๊ฐ€๋ฅด์น˜๋Š” ์˜์–ด ์ˆ˜์—…์„ ๋ณด๊ฒŒ ๋  ๊ฒƒ์ž…๋‹ˆ๋‹ค. ๊ฐ ๋™์˜์ƒ ํŽ˜์ด์ง€์— ํ‘œ์‹œ๋˜๋Š” ์˜์–ด ์ž๋ง‰์„ ๋”๋ธ” ํด๋ฆญํ•˜๋ฉด ๊ทธ๊ณณ์—์„œ ๋™์˜์ƒ์ด ์žฌ์ƒ๋ฉ๋‹ˆ๋‹ค. ๋น„๋””์˜ค ์žฌ์ƒ์— ๋งž์ถฐ ์ž๋ง‰์ด ์Šคํฌ๋กค๋ฉ๋‹ˆ๋‹ค. ์˜๊ฒฌ์ด๋‚˜ ์š”์ฒญ์ด ์žˆ๋Š” ๊ฒฝ์šฐ ์ด ๋ฌธ์˜ ์–‘์‹์„ ์‚ฌ์šฉํ•˜์—ฌ ๋ฌธ์˜ํ•˜์‹ญ์‹œ์˜ค.

https://forms.gle/WvT1wiN1qDtmnspy7