Training artificial intelligence - 6 Minute English

104,891 views ใƒป 2020-03-26

BBC Learning English


์•„๋ž˜ ์˜๋ฌธ์ž๋ง‰์„ ๋”๋ธ”ํด๋ฆญํ•˜์‹œ๋ฉด ์˜์ƒ์ด ์žฌ์ƒ๋ฉ๋‹ˆ๋‹ค. ๋ฒˆ์—ญ๋œ ์ž๋ง‰์€ ๊ธฐ๊ณ„ ๋ฒˆ์—ญ๋ฉ๋‹ˆ๋‹ค.

00:07
Hello. This is 6 Minute English from BBC Learning
0
7960
2840
์•ˆ๋…•ํ•˜์„ธ์š”. BBC Learning English์˜ 6๋ถ„ ์˜์–ด์ž…๋‹ˆ๋‹ค
00:10
English. Iโ€™m Neil.
1
10809
1200
. ์ €๋Š” ๋‹์ž…๋‹ˆ๋‹ค.
00:12
And Iโ€™m Sam.
2
12009
800
00:12
Do you like cooking, Sam? Thereโ€™s a new
3
12809
2421
๊ทธ๋ฆฌ๊ณ  ์ €๋Š” ์ƒ˜์ž…๋‹ˆ๋‹ค.
์š”๋ฆฌ ์ข‹์•„ํ•˜์„ธ์š”, ์ƒ˜?
00:15
recipe Iโ€™ve been trying out - itโ€™s for
4
15230
2321
์ œ๊ฐ€ ์‹œ๋„ํ•˜๊ณ  ์žˆ๋Š” ์ƒˆ๋กœ์šด ๋ ˆ์‹œํ”ผ๊ฐ€ ์žˆ์Šต๋‹ˆ๋‹ค. ๋ฐ”๋กœ
00:17
โ€˜frosted oystersโ€™.
5
17551
1099
'๋ƒ‰๋™ ๊ตด'์ž…๋‹ˆ๋‹ค.
00:18
Frosted oysters?! Soundsโ€ฆ unusual. How
6
18650
5090
๋ƒ‰๋™ ๊ตด?! ์†Œ๋ฆฌ๊ฐ€โ€ฆ ์–ด๋–ป๊ฒŒ
00:23
do you make it?
7
23740
1000
๋งŒ๋“œ๋‚˜์š”?
00:24
Well, take a pound of chicken, then some cubed
8
24740
2769
์Œ, ๋‹ญ๊ณ ๊ธฐ 1ํŒŒ์šด๋“œ์™€ ์ž˜๊ฒŒ ์ฌ
00:27
pork and half a crushed garlic.
9
27509
1760
๋ผ์ง€๊ณ ๊ธฐ, ๋‹ค์ง„ ๋งˆ๋Š˜ ๋ฐ˜๊ฐœ๋ฅผ ๊ฐ€์ ธ๊ฐ€์„ธ์š”.
00:29
Eh? I thought you said it was for โ€˜frosted
10
29269
2731
๋ญ๋ผ๊ณ ? ๋‚˜๋Š” ๋‹น์‹ ์ด ๊ทธ๊ฒƒ์ด
00:32
oystersโ€™, whatever they are.
11
32000
1920
๋ฌด์—‡์ด๋“ ๊ฐ„์— '๋ƒ‰๋™ ๊ตด'์ด๋ผ๊ณ  ๋งํ•œ ์ค„ ์•Œ์•˜์Šต๋‹ˆ๋‹ค.
00:33
Yes, thatโ€™s right. Now heat it up until
12
33930
2350
์˜ˆ, ๋งž์Šต๋‹ˆ๋‹ค. ์ด์ œ ๋“์„ ๋•Œ๊นŒ์ง€ ๊ฐ€์—ดํ•˜๊ณ 
00:36
boiling and serve with custard.
13
36280
1830
์ปค์Šคํ„ฐ๋“œ์™€ ํ•จ๊ป˜ ์ œ๊ณตํ•˜์‹ญ์‹œ์˜ค.
00:38
Ugh, that sounds disgusting! Who on earth
14
38110
2510
์œผ, ์—ญ๊ฒน๊ฒŒ ๋“ค๋ฆฌ๋„ค์š”! ๋„๋Œ€์ฒด ๋ˆ„๊ฐ€
00:40
told you that recipe?
15
40620
1320
๊ทธ ๋ ˆ์‹œํ”ผ๋ฅผ ์•Œ๋ ค์คฌ๋‚˜์š”?
00:41
Itโ€™s not โ€˜whoโ€™ told me, Sam, but โ€˜whatโ€™.
16
41940
2970
Sam, '๋ˆ„๊ฐ€'๊ฐ€ ์•„๋‹ˆ๋ผ '๋ฌด์—‡์„'์ž…๋‹ˆ๋‹ค.
00:44
In fact, that recipe was made by computers
17
44910
2915
์‚ฌ์‹ค ๊ทธ ๋ ˆ์‹œํ”ผ๋Š” ์˜ค๋Š˜ ํ”„๋กœ๊ทธ๋žจ์˜ ์ฃผ์ œ์ธ
00:47
using artificial intelligence, or AI, which
18
47825
2985
์ธ๊ณต์ง€๋Šฅ(AI)์„ ์ด์šฉํ•ด ์ปดํ“จํ„ฐ๊ฐ€ ๋งŒ๋“  ๊ฒƒ์ž…๋‹ˆ๋‹ค
00:50
is the topic of todayโ€™s programme. In real
19
50810
2896
. ์‹ค์ƒํ™œ์—์„œ
00:53
life, AI is making huge progress - from car
20
53706
2964
AI๋Š” ์ฐจ๋Ÿ‰
00:56
satnavs to detecting cancer cells. But as
21
56670
2780
๋‚ด๋น„๊ฒŒ์ด์…˜์—์„œ ์•”์„ธํฌ ํƒ์ง€์— ์ด๋ฅด๊ธฐ๊นŒ์ง€ ์—„์ฒญ๋‚œ ๋ฐœ์ „์„ ์ด๋ฃจ๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค. ํ•˜์ง€๋งŒ
00:59
you can see from that revolting recipe, things
22
59450
3120
๊ทธ ์—ญ๊ฒจ์šด ๋ ˆ์‹œํ”ผ์—์„œ ์•Œ ์ˆ˜ ์žˆ๋“ฏ์ด ์ผ์ด
01:02
donโ€™t always go according to plan.
23
62570
1840
ํ•ญ์ƒ ๊ณ„ํš๋Œ€๋กœ ์ง„ํ–‰๋˜๋Š” ๊ฒƒ์€ ์•„๋‹™๋‹ˆ๋‹ค.
01:04
So, just how intelligent is artificial intelligence?
24
64410
3750
๊ทธ๋ ‡๋‹ค๋ฉด ์ธ๊ณต์ง€๋Šฅ์€ ์–ผ๋งˆ๋‚˜ ์ง€๋Šฅ์ ์ผ๊นŒ์š”?
01:08
I mean, it definitely needs some cooking lessons!
25
68160
3710
ํ™•์‹คํžˆ ์š”๋ฆฌ ์ˆ˜์—…์ด ํ•„์š”ํ•ฉ๋‹ˆ๋‹ค!
01:11
Right. AI is not as intelligent as we tend
26
71870
3310
์˜ค๋ฅธ์ชฝ. AI๋Š” ์šฐ๋ฆฌ๊ฐ€ ์ƒ๊ฐํ•˜๋Š” ๊ฒƒ๋งŒํผ ์ง€๋Šฅ์ ์ด์ง€ ์•Š์Šต๋‹ˆ๋‹ค
01:15
to think. AI programmes use artificial brain
27
75180
2855
. AI ํ”„๋กœ๊ทธ๋žจ์€ ์ธ๊ณต ๋‡Œ
01:18
cells to roughly imitate real brain cell activity,
28
78035
3245
์„ธํฌ๋ฅผ ์ด์šฉํ•ด ์‹ค์ œ ๋‡Œ์„ธํฌ ํ™œ๋™์„ ๋Œ€๋žต์ ์œผ๋กœ ๋ชจ๋ฐฉ
01:21
but theyโ€™re still a long way behind human
29
81280
2687
ํ•˜์ง€๋งŒ, ์—ฌ์ „ํžˆ ์ธ๊ฐ„
01:23
levels of intelligence. And thatโ€™s my quiz
30
83967
2753
์ˆ˜์ค€์˜ ์ง€๋Šฅ์—๋Š” ํ•œ์ฐธ ๋’ค๋–จ์–ด์ ธ ์žˆ๋‹ค. ์ด๊ฒƒ์ด ์ œ ํ€ด์ฆˆ
01:26
question โ€“ in terms of brain cell count,
31
86720
2868
์งˆ๋ฌธ์ž…๋‹ˆ๋‹ค. ๋‡Œ ์„ธํฌ ์ˆ˜ ์ธก๋ฉด์—์„œ
01:29
what level of intelligence is AI currently
32
89588
3012
AI๋Š” ํ˜„์žฌ ์–ด๋–ค ์ˆ˜์ค€์˜ ์ง€๋Šฅ์„
01:32
working at? Is AI as smart as:
33
92600
2780
๋ฐœํœ˜ํ•˜๊ณ  ์žˆ์Šต๋‹ˆ๊นŒ? AI๋Š”
01:35
a) a frog, b) an earthworm or
34
95380
2860
a) ๊ฐœ๊ตฌ๋ฆฌ, b) ์ง€๋ ์ด ๋˜๋Š”
01:38
c) a bumblebee
35
98240
1340
c) ๋•…๋ฒŒ๋งŒํผ ๋˜‘๋˜‘ํ•ฉ๋‹ˆ๊นŒ?
01:39
Well, I donโ€™t think any of those are good
36
99580
3030
์Œ, ์†”์งํžˆ ๋งํ•ด์„œ ๊ทธ๋“ค ์ค‘ ๋ˆ„๊ตฌ๋„ ํ›Œ๋ฅญํ•œ ์š”๋ฆฌ์‚ฌ๋ผ๊ณ  ์ƒ๊ฐํ•˜์ง€ ์•Š์Šต๋‹ˆ๋‹ค
01:42
cooks either, to be honest. Iโ€™ll say c)
37
102610
2364
. ๋‚˜๋Š”
01:44
a bumblebee, because at least they can
38
104974
2606
์ ์–ด๋„
01:47
make honey!
39
107580
1000
๊ฟ€์„ ๋งŒ๋“ค ์ˆ˜ ์žˆ๊ธฐ ๋•Œ๋ฌธ์— c) ๋•…๋ฒŒ์ด๋ผ๊ณ  ๋งํ•  ๊ฒƒ์ž…๋‹ˆ๋‹ค!
01:48
Nice guess, Sam. Weโ€™ll find out the answer
40
108580
1950
์ข‹์€ ์ถ”์ธก์ด์•ผ, ์ƒ˜. ๋‚˜์ค‘์— ๋‹ต์„ ์•Œ์•„ ๋ณด๊ฒ ์Šต๋‹ˆ๋‹ค
01:50
later. But first letโ€™s find out more about
41
110530
2562
. ๊ทธ๋Ÿฌ๋‚˜ ๋จผ์ €
01:53
how AI misunderstandings like the oyster recipe
42
113092
2868
๊ตด ์š”๋ฆฌ๋ฒ•๊ณผ ๊ฐ™์€ AI ์˜คํ•ด๊ฐ€ ์–ด๋–ป๊ฒŒ
01:55
can happen. Janelle Shane is the author of
43
115960
2819
๋ฐœ์ƒํ•  ์ˆ˜ ์žˆ๋Š”์ง€์— ๋Œ€ํ•ด ์ž์„ธํžˆ ์•Œ์•„ ๋ณด๊ฒ ์Šต๋‹ˆ๋‹ค. Janelle Shane์€
01:58
โ€˜You Look Like a Thing and I Love Youโ€™
44
118779
2551
'You Look Like a Thing and I Love You'์˜ ์ €์ž๋กœ
02:01
in which she tells her amusing
45
121330
1869
๊ทธ๋…€์˜ ์žฌ๋ฏธ์žˆ๋Š”
02:03
experiences and bizarre experiments with AI.
46
123199
2741
๊ฒฝํ—˜๊ณผ AI์— ๋Œ€ํ•œ ๊ธฐ์ดํ•œ ์‹คํ—˜์„ ๋“ค๋ ค์ค๋‹ˆ๋‹ค.
02:05
You Look Like a Thing and I Love You โ€“ thatโ€™s
47
125940
3160
You Look Like a Thing and I Love You โ€“
02:09
a strange title for a book, Neil.
48
129100
2200
Neil์ด๋ผ๋Š” ์ฑ…์˜ ์ด์ƒํ•œ ์ œ๋ชฉ์ž…๋‹ˆ๋‹ค.
02:11
Yes. Itโ€™s another example of AI
49
131300
2720
์˜ˆ. AI ์ž˜๋ชป๋œ ์ปค๋ฎค๋‹ˆ์ผ€์ด์…˜์˜ ๋˜ ๋‹ค๋ฅธ ์˜ˆ์ž…๋‹ˆ๋‹ค
02:14
miscommunication.
50
134028
1682
.
02:15
The book title is what a AI produced when
51
135710
2465
์ฑ… ์ œ๋ชฉ์€
02:18
asked to write chat-up lines โ€“ remarks men
52
138175
2525
๋‚จ์„ฑ
02:20
and women make to start up a conversation
53
140700
2351
๊ณผ ์—ฌ์„ฑ์ด ์•Œ์ง€
02:23
with someone they donโ€™t know but find attractive.
54
143051
2809
๋ชปํ•˜์ง€๋งŒ ๋งค๋ ฅ์ ์ด๋ผ๊ณ  โ€‹โ€‹์ƒ๊ฐํ•˜๋Š” ์‚ฌ๋žŒ๊ณผ ๋Œ€ํ™”๋ฅผ ์‹œ์ž‘ํ•˜๊ธฐ ์œ„ํ•ด ํ•˜๋Š” ๋ฐœ์–ธ์„ ์ž‘์„ฑํ•˜๋ผ๋Š” ์š”์ฒญ์„ ๋ฐ›์•˜์„ ๋•Œ AI๊ฐ€ ์ƒ์„ฑํ•œ ๊ฒƒ์ž…๋‹ˆ๋‹ค.
02:25
Here she is talking to the BBC World Service
55
145860
2548
์—ฌ๊ธฐ์—์„œ ๊ทธ๋…€๋Š” BBC World Service
02:28
programme More or Less:
56
148408
1592
ํ”„๋กœ๊ทธ๋žจ More or Less์— ๋Œ€ํ•ด ์ด์•ผ๊ธฐํ•˜๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
02:30
So โ€˜Machine learningโ€™ is what most programmers
57
150000
3830
๊ทธ๋ž˜์„œ '๋จธ์‹  ๋Ÿฌ๋‹'์€ ๋Œ€๋ถ€๋ถ„์˜ ํ”„๋กœ๊ทธ๋ž˜๋จธ๊ฐ€
02:33
mean when they say โ€˜AIโ€™. In the programme
58
153830
3229
'AI'๋ผ๊ณ  ๋งํ•  ๋•Œ ์˜๋ฏธํ•˜๋Š” ๊ฒƒ์ž…๋‹ˆ๋‹ค. ์šฐ๋ฆฌ์—๊ฒŒ ์ต์ˆ™ํ•œ ํ”„๋กœ๊ทธ๋žจ์—์„œ
02:37
that weโ€™re used to, if you want to have
59
157059
3071
02:40
a computer programme solve a problem you have
60
160130
3000
์ปดํ“จํ„ฐ ํ”„๋กœ๊ทธ๋žจ์ด ๋ฌธ์ œ๋ฅผ ํ•ด๊ฒฐํ•˜๋„๋ก ํ•˜๋ ค๋ฉด
02:43
to have a human programmer write down exhaustive
61
163130
3200
์ธ๊ฐ„ ํ”„๋กœ๊ทธ๋ž˜๋จธ๊ฐ€
02:46
step-by-step instructions on how to do everything.
62
166330
2962
๋ชจ๋“  ์ž‘์—…์„ ์ˆ˜ํ–‰ํ•˜๋Š” ๋ฐฉ๋ฒ•์— ๋Œ€ํ•œ ์ฒ ์ €ํ•œ ๋‹จ๊ณ„๋ณ„ ์ง€์นจ์„ ์ž‘์„ฑํ•ด์•ผ ํ•ฉ๋‹ˆ๋‹ค.
02:49
But with โ€˜machine learningโ€™ you just give
63
169292
2428
ํ•˜์ง€๋งŒ '๋จธ์‹  ๋Ÿฌ๋‹'์„ ์‚ฌ์šฉํ•˜๋ฉด ๋ชฉํ‘œ๋ฅผ ๋ถ€์—ฌํ•˜๊ธฐ๋งŒ ํ•˜๋ฉด
02:51
it the goal, and then the programme figures
64
171720
2396
ํ”„๋กœ๊ทธ๋žจ์ด
02:54
out via trial and error how itโ€™s going to
65
174116
2284
์‹œํ–‰์ฐฉ์˜ค๋ฅผ ํ†ตํ•ด
02:56
solve that problem.
66
176400
1440
๋ฌธ์ œ๋ฅผ ํ•ด๊ฒฐํ•˜๋Š” ๋ฐฉ๋ฒ•์„ ์•Œ์•„๋ƒ…๋‹ˆ๋‹ค.
02:57
So even though weโ€™re talking about machines
67
177840
2430
๋”ฐ๋ผ์„œ ์Šค์Šค๋กœ ํ•™์Šตํ•˜๋Š” ๊ธฐ๊ณ„์— ๋Œ€ํ•ด ์ด์•ผ๊ธฐํ•˜๊ณ  ์žˆ์ง€๋งŒ ์—ฌ์ •์„
03:00
learning for themselves, there still need
68
180270
2495
03:02
to be humans involved at the start of the
69
182765
2495
์‹œ์ž‘ํ•  ๋•Œ ์—ฌ์ „ํžˆ ์‚ฌ๋žŒ์ด ์ฐธ์—ฌํ•ด์•ผ ํ•ฉ๋‹ˆ๋‹ค
03:05
journey. This human teaching is done by computer
70
185260
3600
. ์ด ์ธ๊ฐ„ ๊ต์œก์€ ์ปดํ“จํ„ฐ
03:08
programmers โ€“ people who write, or code,
71
188860
3420
ํ”„๋กœ๊ทธ๋ž˜๋จธ, ์ฆ‰
03:12
the computer programmes used by AI.
72
192280
2260
AI๊ฐ€ ์‚ฌ์šฉํ•˜๋Š” ์ปดํ“จํ„ฐ ํ”„๋กœ๊ทธ๋žจ์„ ์ž‘์„ฑํ•˜๊ฑฐ๋‚˜ ์ฝ”๋”ฉํ•˜๋Š” ์‚ฌ๋žŒ๋“ค์ด ์ˆ˜ํ–‰ํ•ฉ๋‹ˆ๋‹ค.
03:14
Right. These programmers write algorithms
73
194540
2520
์˜ค๋ฅธ์ชฝ. ์ด ํ”„๋กœ๊ทธ๋ž˜๋จธ๋Š”
03:17
โ€“ a set of rules or procedures to be followed
74
197060
3041
03:20
in problem-solving exercises. So, for example,
75
200101
3109
๋ฌธ์ œ ํ•ด๊ฒฐ ์—ฐ์Šต์—์„œ ๋”ฐ๋ผ์•ผ ํ•  ์ผ๋ จ์˜ ๊ทœ์น™ ๋˜๋Š” ์ ˆ์ฐจ์ธ ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ์ž‘์„ฑํ•ฉ๋‹ˆ๋‹ค. ์˜ˆ๋ฅผ ๋“ค์–ด
03:23
the AI that wrote that oyster recipe read
76
203210
2602
๊ตด ์š”๋ฆฌ๋ฒ•์„ ์ž‘์„ฑํ•œ AI๋Š” ์ž์ฒด ๋ฒ„์ „์„
03:25
thousands of other recipes before coming up
77
205812
2728
๋งŒ๋“ค๊ธฐ ์ „์— ์ˆ˜์ฒœ ๊ฐœ์˜ ๋‹ค๋ฅธ ์š”๋ฆฌ๋ฒ•์„ ์ฝ์—ˆ์Šต๋‹ˆ๋‹ค
03:28
with its own version.
78
208540
1000
.
03:29
In other words, artificial intelligence uses
79
209540
3040
์ฆ‰, ์ธ๊ณต ์ง€๋Šฅ์€ ๊ฐ€์žฅ ์„ฑ๊ณต์ ์ธ ๋ฐฉ๋ฒ•์„ ์ฐพ์„ ๋•Œ๊นŒ์ง€ ๋™์ผํ•œ ์ž‘์—…์„
03:32
a process of trial and error โ€“ repeating
80
212580
3185
๋ฐ˜๋ณตํ•˜๋Š” ์‹œํ–‰์ฐฉ์˜ค์˜ ๊ณผ์ •์„ ์‚ฌ์šฉํ•ฉ๋‹ˆ๋‹ค
03:35
the same task over and over until finding
81
215765
3265
03:39
the most successful way. Only in the case
82
219030
2730
. ๋‹ค๋งŒ
03:41
of the oyster recipe, there was more โ€˜errorโ€™
83
221760
2300
๊ตด ๋ ˆ์‹œํ”ผ์˜ ๊ฒฝ์šฐ '
03:44
than โ€˜trialโ€™!
84
224060
1220
์‹œํ–‰'๋ณด๋‹ค '์‹ค์ˆ˜'๊ฐ€ ๋” ๋งŽ์•˜๋‹ค!
03:45
Well, according to Janelle Shane, we can learn
85
225280
2680
์Œ, Janelle Shane์— ๋”ฐ๋ฅด๋ฉด, ์šฐ๋ฆฌ๋Š”
03:47
a lot about something by seeing how it
86
227960
2276
๊ทธ๊ฒƒ์ด ์–ด๋–ป๊ฒŒ ์ž˜๋ชป๋˜๋Š”์ง€๋ฅผ ๋ด„์œผ๋กœ์จ ๋ฌด์–ธ๊ฐ€์— ๋Œ€ํ•ด ๋งŽ์€ ๊ฒƒ์„ ๋ฐฐ์šธ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค
03:50
goes wrong. Here she is, talking about an
87
230236
500
03:50
AI which had been told to solve maths problems:
88
230736
4824
. ๊ทธ๋…€๋Š”
์ˆ˜ํ•™ ๋ฌธ์ œ๋ฅผ ํ’€๋„๋ก ์ง€์‹œ๋ฐ›์€ AI์— ๋Œ€ํ•ด ์ด์•ผ๊ธฐํ•˜๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค. ์˜ค๋‹ต์˜ ์ˆ˜์— ๋”ฐ๋ผ
03:55
It seemed to be that it was getting scored
89
235560
3470
์ ์ˆ˜๋ฅผ ๋งค๊ธฐ๋Š” ๊ฒƒ ๊ฐ™์•˜๊ณ 
03:59
on how many wrong answers it got, and it was
90
239030
2447
04:01
supposed to be minimising the number of wrong
91
241477
2503
์˜ค๋‹ต์˜ ์ˆ˜๋ฅผ ์ตœ์†Œํ™”ํ•ด์•ผ ํ–ˆ์Šต๋‹ˆ๋‹ค. ์‹œํ–‰ ์ฐฉ์˜ค์˜
04:03
answers, and just by a stroke of luck as part
92
243980
3720
์ผ๋ถ€๋กœ ์šด์ด ์ข‹์•˜์Šต๋‹ˆ๋‹ค
04:07
of its trial and error flailing around, one
93
247700
2860
.
04:10
of the flails it did accidentally deleted
94
250560
3620
๋„๋ฆฌ๊นจ ์ค‘ ํ•˜๋‚˜๊ฐ€ ์‹ค์ˆ˜๋กœ
04:14
the solutions list, and then it and everybody
95
254180
3029
์†”๋ฃจ์…˜ ๋ชฉ๋ก์„ ์‚ญ์ œ ํ•œ ๋‹ค์Œ ๋‹ค๋ฅธ ๋ชจ๋“  ์‚ฌ๋žŒ๋“ค์ด
04:17
else got a perfect score.
96
257209
2031
๋งŒ์ ์„ ๋ฐ›์•˜์Šต๋‹ˆ๋‹ค.
04:19
So, AIs learn by minimising their errors โ€“ reducing
97
259240
4390
๋”ฐ๋ผ์„œ AI๋Š” ์˜ค๋ฅ˜๋ฅผ ์ตœ์†Œํ™”ํ•˜์—ฌ ํ•™์Šตํ•ฉ๋‹ˆ๋‹ค. ์ฆ‰, ์˜ค๋ฅ˜๋ฅผ
04:23
them as much as possible. And sometimes,
98
263630
2726
์ตœ๋Œ€ํ•œ ์ค„์ž…๋‹ˆ๋‹ค. ๊ทธ๋ฆฌ๊ณ  ๋•Œ๋•Œ๋กœ
04:26
these algorithms only discover the right answer
99
266356
3204
์ด๋Ÿฌํ•œ ์•Œ๊ณ ๋ฆฌ์ฆ˜์€
04:29
by a stroke of luck โ€“ when something unexpected
100
269560
3478
ํ–‰์šด์ด๋‚˜ ์šฐ์—ฐ์— ์˜ํ•ด ์˜ˆ์ƒ์น˜ ๋ชปํ•œ ์ผ์ด
04:33
happens by good luck or chance. It seems to
101
273038
3182
๋ฐœ์ƒํ–ˆ์„ ๋•Œ์—๋งŒ ์ •๋‹ต์„ ๋ฐœ๊ฒฌํ•ฉ๋‹ˆ๋‹ค. ๊ฒฐ๊ตญ
04:36
me that theyโ€™re not so intelligent
102
276220
1860
๊ทธ๋“ค์€ ๊ทธ๋ ‡๊ฒŒ ๋˜‘๋˜‘ํ•˜์ง€ ์•Š์€ ๊ฒƒ ๊ฐ™์Šต๋‹ˆ๋‹ค
04:38
after all!
103
278080
820
04:38
Well, letโ€™s settle it once and for all by
104
278900
2080
!
์ž,
04:40
answering todayโ€™s quiz question.
105
280980
1981
์˜ค๋Š˜์˜ ํ€ด์ฆˆ ๋ฌธ์ œ๋ฅผ ํ’€๋ฉด์„œ ํ•œ ๋ฒˆ์— ํ•ด๊ฒฐํ•ด ๋ด…์‹œ๋‹ค.
04:42
Remember I asked you how intelligent AI was
106
282961
2759
04:45
in terms of brain cell count and you said,
107
285720
2600
๋‡Œ ์„ธํฌ ์ˆ˜ ์ธก๋ฉด์—์„œ ์ธ๊ณต ์ง€๋Šฅ์ด ์–ผ๋งˆ๋‚˜ ์ง€๋Šฅ์ ์ธ์ง€ ๋ฌผ์—ˆ๊ณ  ๋‹น์‹ ์€
04:48
as intelligent as...
108
288320
1760
์ง€๋Šฅ๋งŒํผ ...
04:50
I said c) a bumblebee.
109
290080
1820
c) ๋•…๋ฒŒ์ด๋ผ๊ณ  ๋งํ–ˆ์Šต๋‹ˆ๋‹ค.
04:51
Well, hereโ€™s Janelle again with the answerโ€ฆ
110
291900
2860
๊ธ€์Ž„์š”, Janelle์ด ๋‹ต์„ ๊ฐ€์ง€๊ณ  ๋‹ค์‹œ ์™”์Šต๋‹ˆ๋‹คโ€ฆ
04:54
If youโ€™re looking at rough computing power,
111
294760
3340
๋Œ€๋žต์ ์ธ ์ปดํ“จํŒ… ์„ฑ๋Šฅ์„ ๋ณด๊ณ  ์žˆ๋‹ค๋ฉด
04:58
the algorithms weโ€™re working with are probably
112
298110
2393
์šฐ๋ฆฌ๊ฐ€ ์ž‘์—…ํ•˜๊ณ  ์žˆ๋Š” ์•Œ๊ณ ๋ฆฌ์ฆ˜์€ ์•„๋งˆ๋„
05:00
somewhere around the level of an earthworm.
113
300503
2237
์ง€๋ ์ด ์ˆ˜์ค€ ์ •๋„์ผ ๊ฒƒ์ž…๋‹ˆ๋‹ค.
05:02
So, the correct answer was b) as clever as
114
302740
3910
๊ทธ๋ž˜์„œ ์ •๋‹ต์€ b) ์ง€๋ ์ด๋งŒํผ ์˜๋ฆฌํ•˜๋‹ค
05:06
an earthworm! No wonder AIs canโ€™t cook!
115
306650
3010
! AI๊ฐ€ ์š”๋ฆฌ๋ฅผ ๋ชปํ•˜๋Š” ๊ฒƒ๋„ ๋‹น์—ฐํ•ฉ๋‹ˆ๋‹ค!
05:09
Or take a maths test without cheating! In
116
309660
3110
์•„๋‹ˆ๋ฉด ๋ถ€์ •ํ–‰์œ„ ์—†์ด ์ˆ˜ํ•™ ์‹œํ—˜์„ ์น˜๋ฅด์„ธ์š”!
05:12
this programme weโ€™ve been looking at artificial
117
312770
2360
์ด ํ”„๋กœ๊ทธ๋žจ์—์„œ ์šฐ๋ฆฌ๋Š” ์ธ๊ณต
05:15
intelligence, or AI, and seeing how programmers
118
315130
2360
์ง€๋Šฅ ๋˜๋Š” AI๋ฅผ ์‚ดํŽด๋ณด๊ณ 
05:17
โ€“ thatโ€™s people who write instructions
119
317490
2232
05:19
for computers to follow create algorithms
120
319722
2408
์ปดํ“จํ„ฐ๊ฐ€ ๋”ฐ๋ผ์•ผ ํ•  ์ง€์นจ์„ ์ž‘์„ฑํ•˜๋Š” ํ”„๋กœ๊ทธ๋ž˜๋จธ๊ฐ€ ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ์ƒ์„ฑํ•˜๋Š” ๋ฐฉ๋ฒ•, ์ฆ‰
05:22
โ€“ sets of rules used in problem-solving.
121
322130
2510
๋ฌธ์ œ ํ•ด๊ฒฐ์— ์‚ฌ์šฉ๋˜๋Š” ์ผ๋ จ์˜ ๊ทœ์น™์„ ์‚ดํŽด๋ณด์•˜์Šต๋‹ˆ๋‹ค.
05:24
AI learns through trial and error โ€“ repeating
122
324640
3620
AI๋Š” ์‹œํ–‰์ฐฉ์˜ค๋ฅผ ํ†ตํ•ด ํ•™์Šตํ•ฉ๋‹ˆ๋‹ค. ์ตœ์„ ์˜ ๋ฐฉ๋ฒ•์„
05:28
the same activity again and again until discovering
123
328260
3180
๋ฐœ๊ฒฌํ•  ๋•Œ๊นŒ์ง€ ๋™์ผํ•œ ํ™œ๋™์„ ๋ฐ˜๋ณต
05:31
the best way, and minimising โ€“ reducing
124
331440
3640
ํ•˜๊ณ  ์ตœ์†Œํ™”ํ•˜์—ฌ
05:35
as much as possible, the number of errors
125
335090
2720
๊ฐ€๋Šฅํ•œ ํ•œ ๋งŽ์€ ์˜ค๋ฅ˜ ์ˆ˜๋ฅผ ์ค„์ž…๋‹ˆ๋‹ค
05:37
it makes.
126
337810
1010
.
05:38
And success can be the result of a stroke
127
338820
2340
๊ทธ๋ฆฌ๊ณ  ์„ฑ๊ณต์€ ํ–‰์šด์˜ ๊ฒฐ๊ณผ์ผ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค
05:41
of luck, when something unexpected happens
128
341160
2588
. ์˜ˆ์ƒ์น˜ ๋ชปํ•œ ์ผ์ด ์ˆœ์ „ํžˆ ์šฐ์—ฐํžˆ ๋ฐœ์ƒํ–ˆ์„ ๋•Œ
05:43
purely by chance, although so far that hasnโ€™t
129
343748
2772
, ๋น„๋ก ์ง€๊ธˆ๊นŒ์ง€๋Š”
05:46
helped AIs to write good chat-up lines โ€“ the
130
346520
2393
AI๊ฐ€ ์ข‹์€ ์ฑ„ํŒ… ๋ผ์ธ์„ ์ž‘์„ฑํ•˜๋Š” ๋ฐ ๋„์›€์ด ๋˜์ง€ ์•Š์•˜์ง€๋งŒ,
05:48
flattering remarks people make to get to know
131
348913
2447
์‚ฌ๋žŒ๋“ค์ด
05:51
someone they find attractive.
132
351360
2080
๋งค๋ ฅ์ ์ด๋ผ๊ณ  โ€‹โ€‹์ƒ๊ฐํ•˜๋Š” ์‚ฌ๋žŒ์„ ์•Œ๊ฒŒ ๋˜๊ธฐ ์œ„ํ•ด ํ•˜๋Š” ์•„์ฒจํ•˜๋Š” ๋ง์ž…๋‹ˆ๋‹ค. .
05:53
And AIs donโ€™t know much about cooking oysters
133
353440
2570
๊ทธ๋ฆฌ๊ณ  AI๋„ ๊ตด ์š”๋ฆฌ์— ๋Œ€ํ•ด ์ž˜ ๋ชจ๋ฆ…๋‹ˆ๋‹ค
05:56
either!
134
356010
1000
!
05:57
Thatโ€™s all from us from this programme.
135
357010
1460
๊ทธ๊ฒƒ์ด ์ด ํ”„๋กœ๊ทธ๋žจ์˜ ์ „๋ถ€์ž…๋‹ˆ๋‹ค. BBC ์˜์–ด ํ•™์Šต์„ ์œ„ํ•œ 6๋ถ„ ์˜์–ด์—์„œ
05:58
Be sure to join us again for more topical
136
358470
2188
๋” ๋งŽ์€ ์ฃผ์ œ ํ† ๋ก ๊ณผ ์–ดํœ˜๋ฅผ ์œ„ํ•ด ๋‹ค์‹œ ์ฐธ์—ฌํ•˜์„ธ์š”
06:00
discussion and vocabulary at 6 Minute English
137
360658
2402
06:03
for BBC Learning English. Bye for now!
138
363060
2540
. ์ง€๊ธˆ์€ ์•ˆ๋…•!
06:05
Bye.
139
365600
1300
์•ˆ๋…•.
์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