AI1

1
00:00:16,515 –> 00:00:20,854
-[SONIC BEEPING]
-[FAST BEAT TECHNO MUSIC]

2
00:00:38,437 –> 00:00:41,174
[BEEPING]

3
00:00:41,207 –> 00:00:44,644
[FAST BEAT TENSE MUSIC]

4
00:01:00,126 –> 00:01:02,895
The number one threat

5
00:01:02,928 –> 00:01:07,734
is actually something
humans have created.

6
00:01:12,938 –> 00:01:18,211
NARRATOR: It seems everybody
is talking about
artificial intelligence.

7
00:01:18,244 –> 00:01:23,483
And yet, very few people
actually understand
what it really means.

8
00:01:27,253 –> 00:01:31,491
Very few understand just how
dangerous the conscious mind

9
00:01:31,524 –> 00:01:34,327
of an artificial
intelligence will be.

10
00:01:43,002 –> 00:01:48,274
In a few short years,
mankind will no longer be
the fittest of the species.

11
00:01:52,778 –> 00:01:57,083
We will no longer
have our own future
in our hands.

12
00:01:57,116 –> 00:01:59,853
Free will shall
no longer exist.

13
00:02:05,891 –> 00:02:09,295
Everything we do,
say, and even think

14
00:02:09,328 –> 00:02:11,931
will be monitored
and manipulated

15
00:02:11,964 –> 00:02:16,302
by the most
powerful intelligence
the Earth has ever known.

16
00:02:21,840 –> 00:02:26,779
We will be the cattle,
prodded and probed
by a super being.

17
00:02:26,812 –> 00:02:31,050
And all of this
will only happen,
if we are lucky,

18
00:02:31,083 –> 00:02:35,788
if AI decides that mankind
is so damaging to mother Earth,

19
00:02:35,821 –> 00:02:37,757
then anything could happen.

20
00:02:41,227 –> 00:02:44,363
And all of this will
be unleashed upon us

21
00:02:44,396 –> 00:02:47,700
by mega corporations
out for profit.

22
00:02:50,936 –> 00:02:52,905
Prepare to see the future.

23
00:02:52,938 –> 00:02:55,108
Prepare for
the final countdown.

24
00:02:59,578 –> 00:03:03,182
To truly understand
what is going to happen,

25
00:03:03,215 –> 00:03:07,253
we have to truly understand
artificial intelligence.

26
00:03:07,286 –> 00:03:10,756
It is sometimes called
machine learning.

27
00:03:10,789 –> 00:03:15,428
AI is intelligence
demonstrated by a machine,

28
00:03:15,461 –> 00:03:20,600
not an animal or human,
but a tool, or machine
we have created.

29
00:03:23,936 –> 00:03:26,539
To be truly
artificially intelligent,

30
00:03:26,572 –> 00:03:29,609
the machine has to perceive
its environment

31
00:03:29,642 –> 00:03:34,614
and take actions
to maximize the success
of achieving its goals.

32
00:03:37,816 –> 00:03:41,554
The problem here is that
machines can mimic us.

33
00:03:41,587 –> 00:03:44,523
It can work out how
to get from A to B

34
00:03:44,556 –> 00:03:47,593
by the fastest,
most efficient route.

35
00:03:47,626 –> 00:03:51,664
But this is not intelligence.
This is just logic.

36
00:03:54,800 –> 00:03:57,970
It also has to mirror
the cognitive functions

37
00:03:58,003 –> 00:04:00,940
of the human mind,
such as learning,

38
00:04:00,973 –> 00:04:04,844
problem solving,
and possibly even compassion.

39
00:04:05,911 –> 00:04:08,481
Think about that.
Think about this species

40
00:04:08,514 –> 00:04:10,883
that we’ve essentially
generated,

41
00:04:10,916 –> 00:04:13,386
that has no competition.

42
00:04:13,419 –> 00:04:19,125
They are unrestrained.
And I already say AI is they.

43
00:04:20,459 –> 00:04:22,095
They are not us.

44
00:04:24,029 –> 00:04:26,666
NARRATOR: There are huge
arguments among scientists

45
00:04:26,699 –> 00:04:30,870
as to exactly what constitutes
true artificial intelligence.

46
00:04:35,674 –> 00:04:39,111
Is it optical character
recognition intelligence?

47
00:04:39,144 –> 00:04:42,581
No, it isn’t.
It’s mathematical logic.

48
00:04:42,614 –> 00:04:45,184
Today, people are
saying that a machine

49
00:04:45,217 –> 00:04:47,420
that can understand
human language

50
00:04:47,453 –> 00:04:49,989
is artificially intelligent.

51
00:04:50,022 –> 00:04:54,494
But we have to ask,
what do we mean
by “understand”?

52
00:04:58,864 –> 00:05:01,334
Beating the grandmaster
at chess

53
00:05:01,367 –> 00:05:05,037
is nothing more than a set
of mathematical equations.

54
00:05:05,070 –> 00:05:07,707
In fact, almost
everything that machines

55
00:05:07,740 –> 00:05:11,944
are doing today and termed
artificially intelligent,

56
00:05:11,977 –> 00:05:13,612
are, in fact, nothing more

57
00:05:13,645 –> 00:05:17,317
than super powered logical
computer equations.

58
00:05:22,888 –> 00:05:25,858
In fact, ever since 1956,

59
00:05:25,891 –> 00:05:28,961
when AI became
academic discipline,

60
00:05:28,994 –> 00:05:33,266
there had been arguments,
hopes, and disappointments.

61
00:05:38,504 –> 00:05:40,840
It has been the human inability

62
00:05:40,873 –> 00:05:45,044
to communicate with each other
that has held science back.

63
00:05:45,077 –> 00:05:48,247
Jealousy and greed,
ego and power,

64
00:05:48,280 –> 00:05:51,083
have all conspired
in the mind of man

65
00:05:51,116 –> 00:05:54,921
to hold back his science
fiction utopian ideas.

66
00:06:00,359 –> 00:06:06,199
The goals of AI research
have moved slowly
but definitively forward.

67
00:06:08,567 –> 00:06:12,705
Reasoning, learning,
knowledge building,

68
00:06:12,738 –> 00:06:16,876
planning, language processing,
perception,

69
00:06:16,909 –> 00:06:19,712
and the ability
to utilize robotics

70
00:06:19,745 –> 00:06:22,048
to handle and move objects.

71
00:06:27,219 –> 00:06:29,221
The problem with all of this

72
00:06:29,254 –> 00:06:31,524
is that so many
fields of expertise

73
00:06:31,557 –> 00:06:33,459
have to come together,

74
00:06:33,492 –> 00:06:36,395
computational
mathematical scientists,

75
00:06:36,428 –> 00:06:40,099
computer engineers,
robotics experts,

76
00:06:40,132 –> 00:06:44,137
language experts,
and even philosophers.

77
00:06:47,139 –> 00:06:50,075
All of these have
to be financed,

78
00:06:50,108 –> 00:06:52,711
so we also need marketing men,

79
00:06:52,744 –> 00:06:55,381
economists, accountants,
businessmen,

80
00:06:55,414 –> 00:06:57,283
and so much more.

81
00:06:57,316 –> 00:06:59,285
It is a massive undertaking.

82
00:06:59,318 –> 00:07:02,221
And now is the only
time in history

83
00:07:02,254 –> 00:07:05,425
when all of these things
could truly come together.

84
00:07:11,096 –> 00:07:14,133
Never in human history
has there been

85
00:07:14,166 –> 00:07:17,236
such financial
and even political power

86
00:07:17,269 –> 00:07:20,439
in companies such as
Apple and Microsoft.

87
00:07:20,472 –> 00:07:23,476
Never has there been
a technology capable

88
00:07:23,509 –> 00:07:27,013
on all fronts of
achieving the goal.

89
00:07:31,617 –> 00:07:33,586
Because these kinds
of companies

90
00:07:33,619 –> 00:07:37,490
now have all the money,
power, and expertise

91
00:07:37,523 –> 00:07:41,093
to achieve the goal of true
artificial intelligence,

92
00:07:41,126 –> 00:07:45,698
it means one thing,
AI will be created by,

93
00:07:45,731 –> 00:07:48,234
and controlled by, business.

94
00:07:54,206 –> 00:07:57,676
This means our future world
will be in the hands

95
00:07:57,709 –> 00:08:01,146
of massive, powerful,
and profit-led businessmen

96
00:08:01,179 –> 00:08:04,250
that neither you
nor I have elected.

97
00:08:09,755 –> 00:08:12,658
Our artificially
intelligent future

98
00:08:12,691 –> 00:08:16,061
will be controlled
by the rich elite.

99
00:08:16,094 –> 00:08:17,663
Are you afraid yet?

100
00:08:17,696 –> 00:08:19,065
We haven’t even started.

101
00:08:24,836 –> 00:08:27,373
Many years ago,
scientists claimed

102
00:08:27,406 –> 00:08:31,043
that human intelligence
could be precisely described

103
00:08:31,076 –> 00:08:33,579
in a mathematical structure.

104
00:08:33,611 –> 00:08:36,714
And because of this,
they claimed the machine

105
00:08:36,748 –> 00:08:39,285
would be able to simulate it.

106
00:08:39,318 –> 00:08:42,021
If this is true,
and what if it is,

107
00:08:42,054 –> 00:08:43,255
then we could be creating

108
00:08:43,288 –> 00:08:47,359
a super powerful version
of ourselves.

109
00:08:47,392 –> 00:08:49,595
But it is not entirely true.

110
00:08:54,633 –> 00:08:58,237
What is the true nature
of the human mind?

111
00:08:58,270 –> 00:09:01,407
Is it truly just
a series of equations?

112
00:09:01,440 –> 00:09:03,542
Can the soul be chalked up

113
00:09:03,575 –> 00:09:05,978
on a blackboard by
a mathematician?

114
00:09:06,011 –> 00:09:08,781
Indeed, is there even a soul?

115
00:09:13,151 –> 00:09:16,155
Have we, in fact,
been lying to ourselves

116
00:09:16,188 –> 00:09:18,324
as a safety device
through fear

117
00:09:18,357 –> 00:09:20,893
of feeling alone
in the universe?

118
00:09:20,926 –> 00:09:23,729
Are we brave enough,
or stupid enough,

119
00:09:23,762 –> 00:09:25,898
to unleash something
we cannot yet

120
00:09:25,931 –> 00:09:28,100
truly understand ourselves?

121
00:09:33,872 –> 00:09:37,309
The fact is, we are closer
to unleashing the quantum

122
00:09:37,342 –> 00:09:40,946
computer-powered super
artificially intelligent mind

123
00:09:40,979 –> 00:09:45,017
upon the world than we
are being led to believe.

124
00:09:45,050 –> 00:09:48,921
And it will cause
a greater revolution
upon the globe

125
00:09:48,954 –> 00:09:51,157
than human history
has ever known.

126
00:09:52,691 –> 00:09:54,527
If you think about it,

127
00:09:55,394 –> 00:09:58,697
humankind on the whole

128
00:09:58,730 –> 00:10:01,133
is no longer threatened

129
00:10:01,166 –> 00:10:04,804
by things that have
historically threatened us.

130
00:10:06,405 –> 00:10:08,708
Animals, the weather,

131
00:10:10,375 –> 00:10:13,379
human shortages of food,

132
00:10:13,412 –> 00:10:17,549
crops affected by
the season, disease,

133
00:10:17,582 –> 00:10:22,154
all of these things in
the last few centuries

134
00:10:22,187 –> 00:10:25,324
have not been conquered,
but I will say

135
00:10:25,357 –> 00:10:27,793
mitigated to a point
that they are no longer

136
00:10:27,826 –> 00:10:31,063
a threat to our population
and our existence.

137
00:10:33,398 –> 00:10:35,034
NARRATOR:
The Industrial Revolution

138
00:10:35,067 –> 00:10:37,002
will pale into insignificance

139
00:10:37,035 –> 00:10:41,006
compared to the radical changes
that will ensue.

140
00:10:41,039 –> 00:10:45,310
From health care to warfare,
from movies to games,

141
00:10:45,343 –> 00:10:47,479
everything will be changed,

142
00:10:47,512 –> 00:10:50,816
and massive unemployment
and social unrest

143
00:10:50,849 –> 00:10:52,585
will be the inevitable result.

144
00:11:00,092 –> 00:11:03,328
Even now, governments
are testing the process

145
00:11:03,361 –> 00:11:05,397
of universal payments.

146
00:11:05,430 –> 00:11:07,733
These experiments
whereby people

147
00:11:07,766 –> 00:11:10,436
are being paid to do nothing

148
00:11:10,469 –> 00:11:12,771
are costing millions
of dollars.

149
00:11:12,804 –> 00:11:16,442
And so far, they have not
revealed good results.

150
00:11:18,443 –> 00:11:21,747
People do not like
to do nothing.

151
00:11:26,251 –> 00:11:30,189
The massive increase in
computer technology and power,

152
00:11:30,222 –> 00:11:34,193
alongside the relatively
cheap cost of memory,

153
00:11:34,226 –> 00:11:38,397
have resulted in
unprecedented amounts of data.

154
00:11:38,430 –> 00:11:41,066
This, in turn,
has aided in the learning

155
00:11:41,099 –> 00:11:43,268
of knowledge by the machines.

156
00:11:43,301 –> 00:11:46,371
They are learning
everything about us

157
00:11:46,404 –> 00:11:47,740
and keeping the data.

158
00:11:53,411 –> 00:11:58,217
Everything you and I do online,
and quite often offline,

159
00:11:58,250 –> 00:12:02,387
is being monitored,
stored, and learned from.

160
00:12:02,420 –> 00:12:05,624
Machine learning is
now so sophisticated

161
00:12:05,657 –> 00:12:08,393
that it is even writing
its own programs

162
00:12:08,426 –> 00:12:10,162
and computer language.

163
00:12:10,195 –> 00:12:12,998
Indeed, in order
to better work,

164
00:12:13,031 –> 00:12:15,601
one program created
its own language

165
00:12:15,634 –> 00:12:18,104
and hid it
from human overseers.

166
00:12:24,276 –> 00:12:29,248
In 2011,
IBM’s AI machine, Watson,

167
00:12:29,281 –> 00:12:32,017
played the Jeopardy
TV quiz show,

168
00:12:32,050 –> 00:12:34,453
and defeated the two
greatest champions

169
00:12:34,486 –> 00:12:36,455
the show had ever known.

170
00:12:36,488 –> 00:12:38,824
It didn’t win by
a small margin.

171
00:12:38,857 –> 00:12:40,192
It destroyed them.

172
00:12:44,863 –> 00:12:49,001
Humans were no match for Watson
all those years ago.

173
00:12:49,034 –> 00:12:52,704
Since then, Watson has had
years of learning,

174
00:12:52,737 –> 00:12:55,007
building up its knowledge base.

175
00:12:55,040 –> 00:12:57,876
Now, there are faster
and faster computers

176
00:12:57,909 –> 00:12:59,945
with more and more data.

177
00:12:59,978 –> 00:13:02,848
The algorithms have
been vastly improved

178
00:13:02,881 –> 00:13:04,917
by man and machine.

179
00:13:09,321 –> 00:13:11,690
Do we really believe
luck plays a part

180
00:13:11,723 –> 00:13:13,625
in those gambling games?

181
00:13:13,658 –> 00:13:16,862
AI technology makes
sure you cannot win.

182
00:13:19,497 –> 00:13:21,967
Today, millions of us interact

183
00:13:22,000 –> 00:13:26,238
with so-called AI technology
on a daily basis.

184
00:13:26,271 –> 00:13:30,242
Alexa, and other
human interfaces
on smart devices

185
00:13:30,275 –> 00:13:32,744
such as pads
and mobile phones,

186
00:13:32,777 –> 00:13:35,080
are all using this information.

187
00:13:39,784 –> 00:13:44,489
Computer games now use AI
and every time we play,

188
00:13:44,522 –> 00:13:47,359
we feed more data
into the machines.

189
00:13:53,098 –> 00:13:56,001
There is a game called
Go, and the champion

190
00:13:56,034 –> 00:13:58,503
was a human called Ke Jie.

191
00:13:58,536 –> 00:14:02,174
He lost the championship
to a machine.

192
00:14:02,207 –> 00:14:06,111
Go is a game
that is far more complex
than chess,

193
00:14:06,144 –> 00:14:09,348
and relies on more
than just equations.

194
00:14:09,381 –> 00:14:13,085
It needs reasoning
and imagination.

195
00:14:17,389 –> 00:14:21,393
In 2012, AI technology
was sporadic.

196
00:14:21,426 –> 00:14:24,596
It was occasionally
let loose on the Internet.

197
00:14:24,629 –> 00:14:28,467
Now, there are thousands
of AI projects out there,

198
00:14:28,500 –> 00:14:31,637
working permanently,
collecting information,

199
00:14:31,670 –> 00:14:35,674
and learning at rates
that humans simply
could not comprehend.

200
00:14:41,146 –> 00:14:43,982
The increase is partly
down to the cheaper

201
00:14:44,015 –> 00:14:47,452
neural networks
available for testing.

202
00:14:47,485 –> 00:14:50,656
As Cloud computing
starts to take over,

203
00:14:50,689 –> 00:14:54,159
this will increase at
a much faster rate.

204
00:14:54,192 –> 00:14:58,263
The systems have learned
human languages so well

205
00:14:58,296 –> 00:15:02,567
that we can now buy ear buds
that automatically translate

206
00:15:02,600 –> 00:15:06,338
the spoken word
into our ears instantly.

207
00:15:12,544 –> 00:15:15,781
There will be no need
again for translators.

208
00:15:15,814 –> 00:15:18,150
Thousands of jobs
will be gone

209
00:15:18,183 –> 00:15:21,086
with just this
one simple device.

210
00:15:21,119 –> 00:15:25,757
Star Trek’s
universal translator
is coming true.

211
00:15:25,790 –> 00:15:28,460
[FIREWORKS WHOOSHING, BLASTING]

212
00:15:30,528 –> 00:15:32,631
One in five companies now state

213
00:15:32,664 –> 00:15:35,367
that they have introduced
AI learning devices

214
00:15:35,400 –> 00:15:39,271
within their organizations,
and this is increasing.

215
00:15:39,304 –> 00:15:42,841
China has invested billions
into AI technology,

216
00:15:42,874 –> 00:15:46,979
and intends to become
the first AI superpower.

217
00:15:53,051 –> 00:15:55,387
AI involves
algorithms that learn

218
00:15:55,420 –> 00:15:57,489
from the data it seeks out.

219
00:15:57,522 –> 00:16:00,492
Strategies, rules, patterns,

220
00:16:00,525 –> 00:16:02,361
and more are all gathered.

221
00:16:06,064 –> 00:16:09,468
Through this collecting
of data, AI systems

222
00:16:09,501 –> 00:16:13,071
will be able to learn
any function it wishes,

223
00:16:13,104 –> 00:16:16,241
whether chess, Go,
or war games.

224
00:16:22,847 –> 00:16:26,218
The learning process
knows no bounds.

225
00:16:26,251 –> 00:16:30,422
All knowledge
is within the grasp
of this super brain.

226
00:16:30,455 –> 00:16:34,426
Everything ever known,
and with knowledge
of our patterns,

227
00:16:34,459 –> 00:16:37,362
strategies, and rules,
it will be able

228
00:16:37,395 –> 00:16:41,166
to instantly predict
the outcome of anything.

229
00:16:46,938 –> 00:16:49,975
This is why it can
beat us at games.

230
00:16:50,008 –> 00:16:54,312
If it destroys all
at the most complex
of human games,

231
00:16:54,345 –> 00:16:56,915
then it can beat us
in anything.

232
00:17:03,188 –> 00:17:07,092
AI has come a long way
from telling the small robot

233
00:17:07,125 –> 00:17:10,129
how to get from A to B,
through an obstacle course.

234
00:17:16,334 –> 00:17:20,072
At the moment, AI lacks
certain human features,

235
00:17:20,105 –> 00:17:24,276
such as that elusive thing
we call common sense.

236
00:17:24,309 –> 00:17:29,247
Our reasoning skills are
almost unconscious to us.

237
00:17:29,280 –> 00:17:32,284
We inhabit
three-dimensional space,

238
00:17:32,317 –> 00:17:34,152
and without even thinking,

239
00:17:34,185 –> 00:17:38,323
we know the distance
from us to this or that.

240
00:17:38,356 –> 00:17:42,594
We listen for threats without
being aware that we are.

241
00:17:42,627 –> 00:17:46,698
The AI machine has to be
aware of all these things,

242
00:17:46,731 –> 00:17:50,169
and presently does so
in a mathematical way.

243
00:17:55,907 –> 00:17:57,642
A child of two knows

244
00:17:57,675 –> 00:17:59,845
that if he pushes
a cork off the table,

245
00:17:59,878 –> 00:18:01,413
it will fall to the ground.

246
00:18:02,480 –> 00:18:05,183
AI has to compute
the whole thing

247
00:18:05,216 –> 00:18:08,153
through a complex
series of algorithms.

248
00:18:13,391 –> 00:18:16,228
There are many more things
that we naturally learn

249
00:18:16,261 –> 00:18:18,563
without even realizing it,

250
00:18:18,596 –> 00:18:21,366
such as the ability
to speak to others

251
00:18:21,399 –> 00:18:23,902
in a way that
they can understand.

252
00:18:23,935 –> 00:18:28,206
AI has had a really sharp
learning curve in this area,

253
00:18:28,239 –> 00:18:31,843
and translating language
for us is very different

254
00:18:31,876 –> 00:18:34,913
to actually understanding
what is being said,

255
00:18:34,946 –> 00:18:36,181
and what is meant.

256
00:18:41,386 –> 00:18:45,323
Quite often, we will use
analogies and double meanings.

257
00:18:45,356 –> 00:18:48,193
AI would presently
have a lot of trouble

258
00:18:48,226 –> 00:18:50,195
understanding such things.

259
00:18:50,228 –> 00:18:53,598
However, the more
it interacts with us,

260
00:18:53,631 –> 00:18:56,635
the more it learns
our strange ways.

261
00:19:03,441 –> 00:19:04,910
This is one of the reasons

262
00:19:04,943 –> 00:19:08,413
that driverless cars
can make mistakes.

263
00:19:08,446 –> 00:19:11,850
The AI technology
fails to comprehend

264
00:19:11,883 –> 00:19:15,954
what other drivers
or humans are going to do.

265
00:19:21,659 –> 00:19:26,731
It makes decisions
that are not the same
as human decisions.

266
00:19:26,764 –> 00:19:31,236
We make them intuitively,
and it makes them using logic.

267
00:19:35,240 –> 00:19:38,443
It’s like trying to
explain joy to Spock.

268
00:19:43,314 –> 00:19:47,385
And all of this means
that there are problems
ahead for us.

269
00:19:47,418 –> 00:19:49,588
When AI becomes conscious,

270
00:19:49,621 –> 00:19:52,424
when it rears its machine head,

271
00:19:52,457 –> 00:19:54,426
will it understand us,

272
00:19:54,459 –> 00:19:57,796
will it know what
being alive means,

273
00:19:57,829 –> 00:20:01,433
the joy, the fear,
the love, the hate?

274
00:20:07,338 –> 00:20:11,076
How can it comprehend
these things in us

275
00:20:11,109 –> 00:20:13,011
and know compassion?

276
00:20:15,513 –> 00:20:17,782
Can it comprehend compassion?

277
00:20:17,815 –> 00:20:19,618
Or will it simply see us

278
00:20:19,651 –> 00:20:23,622
as pathetic biological forms
and destroy us?

279
00:20:25,056 –> 00:20:27,425
Think about the implications

280
00:20:27,458 –> 00:20:32,497
of a race,
an animal species,
if you will,

281
00:20:32,530 –> 00:20:36,068
that is faster, more efficient,

282
00:20:37,068 –> 00:20:41,673
is replacing people
in their work places,

283
00:20:42,473 –> 00:20:46,011
and has no need, no desire,

284
00:20:46,044 –> 00:20:52,183
no call to sustenance,

285
00:20:52,216 –> 00:20:54,352
really beyond electricity,

286
00:20:54,385 –> 00:20:57,689
which humans kind of
also need now to survive.

287
00:20:59,824 –> 00:21:02,394
NARRATOR: AI technology
relies heavily

288
00:21:02,427 –> 00:21:05,230
on a thing called
machine learning.

289
00:21:05,263 –> 00:21:09,334
This is when a machine
is set to learn by itself

290
00:21:09,367 –> 00:21:11,670
without supervision or input.

291
00:21:15,673 –> 00:21:18,677
AI is let loose in various ways

292
00:21:18,710 –> 00:21:20,879
to learn from its environment,

293
00:21:20,912 –> 00:21:23,415
whether this
is a computer game,

294
00:21:23,448 –> 00:21:28,453
the Internet, social media,
or even surveillance cameras.

295
00:21:34,559 –> 00:21:36,494
Through the use of algorithms,

296
00:21:36,527 –> 00:21:40,265
the AI program learns
to constantly improve

297
00:21:40,298 –> 00:21:42,267
the inputs and experience.

298
00:21:45,937 –> 00:21:50,408
Without a human present
to constantly check the inputs,

299
00:21:50,441 –> 00:21:54,212
and label the experiences
for the AI program,

300
00:21:54,245 –> 00:21:56,448
it learns without guidance.

301
00:22:02,053 –> 00:22:05,623
Billions of gigabytes
of data are now stored

302
00:22:05,656 –> 00:22:09,961
within these AI programs
of self-learned experience.

303
00:22:09,994 –> 00:22:13,431
Nobody knows what
the AI program knows,

304
00:22:13,464 –> 00:22:15,000
for nobody can.

305
00:22:20,938 –> 00:22:23,742
We simply do not have
the mental capacity,

306
00:22:23,775 –> 00:22:27,545
even now,
to know what AI knows,

307
00:22:27,578 –> 00:22:29,948
let alone, in the future.

308
00:22:35,520 –> 00:22:39,391
In truth, we only have
to look at human history

309
00:22:39,424 –> 00:22:42,727
to know that all
will not be well.

310
00:22:42,760 –> 00:22:47,499
When we split the atom,
we discovered a possible
free energy source.

311
00:22:47,532 –> 00:22:50,402
But mankind took it
and abused it

312
00:22:50,435 –> 00:22:52,704
by creating the nuclear bomb.

313
00:22:58,342 –> 00:23:00,812
This is just one example.

314
00:23:00,845 –> 00:23:04,616
AI may be truly knowledgeable
and super intelligent,

315
00:23:04,649 –> 00:23:07,652
but stupid humans
have created it,

316
00:23:07,685 –> 00:23:09,654
and it is learning from us.

317
00:23:15,693 –> 00:23:17,862
Will it try
to acquire resources

318
00:23:17,895 –> 00:23:20,098
and protect itself from us?

319
00:23:20,131 –> 00:23:22,033
We have to be very careful

320
00:23:22,066 –> 00:23:24,870
what we ask of this
new technology.

321
00:23:27,805 –> 00:23:30,375
Through natural
language processing,

322
00:23:30,408 –> 00:23:33,011
in addition
to the data learned,

323
00:23:33,044 –> 00:23:36,714
AI is now starting
to understand

324
00:23:36,747 –> 00:23:39,818
what is being said
and done by humans.

325
00:23:39,851 –> 00:23:43,354
It reads the news,
listens to the radio,

326
00:23:43,387 –> 00:23:48,059
podcasts, films,
our buying
and selling activities,

327
00:23:48,092 –> 00:23:52,030
our politics, our feelings,
loves and hates.

328
00:23:56,601 –> 00:23:59,003
It is learning
from a civilization

329
00:23:59,036 –> 00:24:00,839
that is full of monsters.

330
00:24:00,872 –> 00:24:03,041
And so we should
not be surprised

331
00:24:03,074 –> 00:24:08,146
if we discover that AI
becomes a monster itself,

332
00:24:08,179 –> 00:24:11,082
maybe not openly
and clear to see,

333
00:24:11,115 –> 00:24:14,052
but in subtle ways
to begin with.

334
00:24:14,085 –> 00:24:18,723
To humanity as it exists now,

335
00:24:19,557 –> 00:24:24,662
I feel that safety,
precaution,

336
00:24:24,695 –> 00:24:28,833
and thought
in what we are doing

337
00:24:28,866 –> 00:24:30,902
with the things that
we are creating,

338
00:24:30,935 –> 00:24:33,671
and how we implement
them in our society,

339
00:24:33,704 –> 00:24:35,673
are some of
the biggest questions

340
00:24:35,706 –> 00:24:37,809
we are not asking.

341
00:24:39,243 –> 00:24:41,045
NARRATOR: Already,
government agencies

342
00:24:41,078 –> 00:24:45,717
and secret service agencies,
are using AI technology

343
00:24:45,750 –> 00:24:47,986
to find terrorists
and criminals.

344
00:24:53,157 –> 00:24:55,460
Marketing companies
are using it

345
00:24:55,493 –> 00:24:58,263
to monitor
mankind’s activities,

346
00:24:58,296 –> 00:25:01,099
to learn
and to influence us.

347
00:25:04,835 –> 00:25:07,171
We do not know it,
but today,

348
00:25:07,204 –> 00:25:08,640
whenever we go online,

349
00:25:08,673 –> 00:25:12,443
we are dealing with
powerful AI technologies.

350
00:25:12,476 –> 00:25:15,179
Our YouTube viewing preferences

351
00:25:15,212 –> 00:25:16,948
are telling marketing companies

352
00:25:16,981 –> 00:25:19,117
what commercials to show us,

353
00:25:19,150 –> 00:25:21,619
and what products
we would most likely

354
00:25:21,652 –> 00:25:23,421
be interested in buying.

355
00:25:23,454 –> 00:25:26,991
This information spreads
across the AI world

356
00:25:27,024 –> 00:25:30,161
and soon, we are wondering
how Facebook knew

357
00:25:30,194 –> 00:25:33,164
we’ve been looking
at that particular product
on Amazon.

358
00:25:38,869 –> 00:25:43,007
It knows because we gave
it permission to spy on us.

359
00:25:43,040 –> 00:25:46,210
These permissions are
all in small print

360
00:25:46,243 –> 00:25:47,812
of the terms and conditions

361
00:25:47,845 –> 00:25:50,682
we all just click through
when signing up.

362
00:25:50,715 –> 00:25:52,017
Nobody reads them.

363
00:25:54,986 –> 00:25:57,322
With huge amounts
of data learning,

364
00:25:57,355 –> 00:26:00,358
how the human race
goes about its business,

365
00:26:00,391 –> 00:26:04,062
machine learning is
coming to a peak point

366
00:26:04,095 –> 00:26:06,831
where it will
suddenly understand.

367
00:26:06,864 –> 00:26:08,232
This happens in nature,

368
00:26:08,265 –> 00:26:10,869
and it will happen
with the machine.

369
00:26:14,038 –> 00:26:17,208
The machine perception
comes from new resources,

370
00:26:17,241 –> 00:26:19,911
the cameras and microphones
on your devices

371
00:26:19,944 –> 00:26:23,181
that you gave permission
to be used to spy on you.

372
00:26:23,214 –> 00:26:24,215
Yes, you did.

373
00:26:26,751 –> 00:26:29,887
The direct keyboard input
you gave on your PCs,

374
00:26:29,920 –> 00:26:33,925
Macs, mobile phones and pads,
but more than this,

375
00:26:33,958 –> 00:26:37,228
it also listens to
wireless information.

376
00:26:37,261 –> 00:26:40,231
Sonar, radar, cameras,
in the streets

377
00:26:40,264 –> 00:26:43,234
and down the store,
using speech recognition,

378
00:26:43,267 –> 00:26:45,269
it knows what’s been said.

379
00:26:45,302 –> 00:26:48,940
Using facial recognition,
it knows who is saying it,

380
00:26:48,973 –> 00:26:52,243
and it can even
read your lips
in any language.

381
00:26:52,276 –> 00:26:54,479
You cannot escape it anywhere.

382
00:27:00,117 –> 00:27:03,688
All of this sounds
a little Big Brother.

383
00:27:03,721 –> 00:27:05,857
It was predicted
many years ago

384
00:27:05,890 –> 00:27:08,426
by a lot of science fiction
authors

385
00:27:08,459 –> 00:27:11,529
who had the foresight
to not necessarily see

386
00:27:11,562 –> 00:27:14,766
what technology will become,
but instead,

387
00:27:14,799 –> 00:27:17,302
knew what humanity
was capable of.

388
00:27:23,007 –> 00:27:26,544
They knew that those in power
would use technology

389
00:27:26,577 –> 00:27:29,313
to keep and grow
their power base.

390
00:27:29,346 –> 00:27:30,982
They always have.

391
00:27:35,920 –> 00:27:39,724
In Medieval Europe,
the church used
the new technology

392
00:27:39,757 –> 00:27:42,093
of stained glass
windows in churches

393
00:27:42,126 –> 00:27:45,830
to put their message across
to the illiterate people,

394
00:27:45,863 –> 00:27:48,733
manipulating them
through colorful images

395
00:27:48,766 –> 00:27:52,470
that told a simple story
of submission to
the mother church.

396
00:27:52,503 –> 00:27:57,608
They added the subtle
and clever layer of repetitive
and memorable music.

397
00:27:57,641 –> 00:28:00,445
The people chanted the message
to themselves,

398
00:28:00,478 –> 00:28:02,047
and were brainwashed.

399
00:28:07,017 –> 00:28:08,920
The glowing screens
of our phones,

400
00:28:08,953 –> 00:28:12,156
pads and laptops
today do the same.

401
00:28:12,189 –> 00:28:14,158
The sheep enter the fold

402
00:28:14,191 –> 00:28:16,294
for the great shepherd
to feed them

403
00:28:16,327 –> 00:28:18,063
before taking them
to the slaughter.

404
00:28:24,301 –> 00:28:27,004
But all of this is
in the cyber world,

405
00:28:27,037 –> 00:28:32,276
the Internet, social media,
AI-created videos and more.

406
00:28:32,309 –> 00:28:35,847
With robotics, we will be
bringing this powerful mind

407
00:28:35,880 –> 00:28:39,084
to very real,
three-dimensional life.

408
00:28:43,087 –> 00:28:47,391
Robotics would stagnate
without the software of AI.

409
00:28:47,424 –> 00:28:50,027
The sheer level of
information needed

410
00:28:50,060 –> 00:28:52,797
to understand
and make movement

411
00:28:52,830 –> 00:28:56,901
was too much before,
now it is a powerful tool

412
00:28:56,934 –> 00:29:00,004
to make movement more
natural and purposeful.

413
00:29:00,037 –> 00:29:03,241
[TENSE TECHNO MUSIC]

414
00:29:04,942 –> 00:29:09,547
Robots are becoming
super efficient
through the software.

415
00:29:09,580 –> 00:29:13,017
Robots now instantly learn
about their environment,

416
00:29:13,050 –> 00:29:15,419
everything from
temperature to obstacles,

417
00:29:15,452 –> 00:29:17,388
from light to sound.

418
00:29:17,421 –> 00:29:21,325
It recognizes people through
facial and aural recognition

419
00:29:21,358 –> 00:29:24,162
and instantly knows
everything about them,

420
00:29:24,195 –> 00:29:26,397
from its vast database.

421
00:29:32,403 –> 00:29:37,408
Think RoboCop, but much,
much faster and powerful.

422
00:29:42,213 –> 00:29:43,948
And then imagine
the next step,

423
00:29:43,981 –> 00:29:48,519
when robots are no longer
mechanical, but biological.

424
00:29:48,552 –> 00:29:52,290
This is happening,
organic robots,

425
00:29:52,323 –> 00:29:55,626
created in the likeness
of mankind,

426
00:29:55,659 –> 00:29:58,830
but faster, smarter,
and stronger.

427
00:30:03,000 –> 00:30:05,469
We will not be able to compete.

428
00:30:05,502 –> 00:30:09,440
And organic robots
will be able to replicate.

429
00:30:12,643 –> 00:30:15,980
To imagine that
we will be able
to somehow place

430
00:30:16,013 –> 00:30:19,550
a genetic safety device
in them is foolish,

431
00:30:19,583 –> 00:30:22,386
for their minds
will be so smart,

432
00:30:22,419 –> 00:30:25,023
they will simply
rewrite the code.

433
00:30:28,993 –> 00:30:32,063
As fun and interesting
robots are,

434
00:30:32,096 –> 00:30:36,067
there are other ways
of bringing AI
into our reality.

435
00:30:36,100 –> 00:30:38,636
Robots make things,
move things,

436
00:30:38,669 –> 00:30:41,505
and do work that
is labor-intensive

437
00:30:41,538 –> 00:30:43,307
and often boring.

438
00:30:43,340 –> 00:30:46,010
They will fill factories
now across the world

439
00:30:46,043 –> 00:30:48,746
and have taken
millions of jobs.

440
00:30:48,779 –> 00:30:52,149
They stand in shopping centers
and pass on information.

441
00:30:52,182 –> 00:30:54,485
They even flip burgers.

442
00:30:54,518 –> 00:30:55,920
But scientists are working

443
00:30:55,953 –> 00:30:59,156
on exoskeletons
for humans, too.

444
00:30:59,189 –> 00:31:02,827
These are robotic skins
that people, such as soldiers,

445
00:31:02,860 –> 00:31:05,129
will wear to make
them super human.

446
00:31:08,465 –> 00:31:10,868
With the addition
of AI technology,

447
00:31:10,901 –> 00:31:14,372
the soldier will make instant
and effective decisions.

448
00:31:18,409 –> 00:31:21,846
Soon, we will be
implanted with chips,

449
00:31:21,879 –> 00:31:25,383
so that we, too,
can access the AI technology.

450
00:31:25,416 –> 00:31:28,386
We will suddenly be
able to know everything,

451
00:31:28,419 –> 00:31:30,855
to access
the Worldwide Web of data

452
00:31:30,888 –> 00:31:33,725
and to pass out information
back into the collective.

453
00:31:38,362 –> 00:31:41,165
Those who do not wish
to be implanted

454
00:31:41,198 –> 00:31:44,969
will be left behind
in the rat race
we call humanity.

455
00:31:45,002 –> 00:31:47,738
We will be outcasts
in a hybrid world

456
00:31:47,771 –> 00:31:50,174
full of AI-assisted humans,

457
00:31:50,207 –> 00:31:52,343
who will be also more likely

458
00:31:52,376 –> 00:31:56,347
to have super improved
limbs, sight and ears.

459
00:31:56,380 –> 00:31:59,183
And we will be convinced
that all of this

460
00:31:59,216 –> 00:32:02,219
is good for the evolution
of the human race,

461
00:32:02,252 –> 00:32:05,690
that a collective mind will
bring peace and harmony.

462
00:32:05,723 –> 00:32:07,258
Don’t be fooled.

463
00:32:12,129 –> 00:32:14,565
Modern computing
has come a long way

464
00:32:14,598 –> 00:32:16,400
from a simple laptop.

465
00:32:16,433 –> 00:32:18,769
Today, the most
powerful computers

466
00:32:18,802 –> 00:32:22,506
are using quantum
processes and neural nets.

467
00:32:22,539 –> 00:32:25,443
These were inspired by
the actual architecture

468
00:32:25,476 –> 00:32:27,611
of the brain’s neural nets,

469
00:32:27,644 –> 00:32:30,148
or the neurons in ours brains.

470
00:32:35,486 –> 00:32:40,424
Modern neural nets learn
continuously while functioning.

471
00:32:40,457 –> 00:32:43,694
It doesn’t have to do
things at different paces.

472
00:32:43,727 –> 00:32:47,365
Today, these neural nets
are used extensively

473
00:32:47,398 –> 00:32:49,400
to predict the stock market.

474
00:32:49,433 –> 00:32:52,703
And they can even do
buying and selling.

475
00:32:52,736 –> 00:32:56,374
They do this much more
successfully than humans.

476
00:32:56,407 –> 00:32:59,110
Because of the success
in these systems,

477
00:32:59,143 –> 00:33:03,314
business has, and is,
investing heavily in them.

478
00:33:03,347 –> 00:33:05,649
And we are seeing
an exponential increase

479
00:33:05,682 –> 00:33:07,752
in their scope and power.

480
00:33:07,785 –> 00:33:11,622
Today, we spend 30 times
more on AI technology

481
00:33:11,655 –> 00:33:14,459
than we did in 2015.

482
00:33:20,597 –> 00:33:24,769
Such technology,
along with software
for deep learning,

483
00:33:24,802 –> 00:33:27,571
is changing the world,
and most people

484
00:33:27,604 –> 00:33:29,674
are blissfully unaware of it.

485
00:33:31,208 –> 00:33:34,745
Something that isn’t
just affecting the individual

486
00:33:34,778 –> 00:33:38,315
with power in their pocket,

487
00:33:38,348 –> 00:33:39,917
amazing computer capabilities

488
00:33:39,950 –> 00:33:42,453
that we didn’t even
have as a society

489
00:33:42,486 –> 00:33:45,189
just a mere number
of years ago,

490
00:33:45,222 –> 00:33:50,828
now each individual person
and child is adapted completely

491
00:33:50,861 –> 00:33:55,833
into the artificial intelligence
in your pocket,

492
00:33:55,866 –> 00:34:00,004
and on a global scale
phenomenon.

493
00:34:00,037 –> 00:34:02,540
Sure, we’re around
technology everyday,

494
00:34:02,573 –> 00:34:07,144
and may not realize its
fast yet integration

495
00:34:07,177 –> 00:34:11,215
into our lives that we’ve come
to just accept as the norm.

496
00:34:15,585 –> 00:34:19,222
NARRATOR:
This slow encroachment
in our everyday lives

497
00:34:19,255 –> 00:34:22,158
goes relatively unnoticed.

498
00:34:22,192 –> 00:34:25,162
It has become a general
purpose technology,

499
00:34:25,195 –> 00:34:26,831
like electricity.

500
00:34:26,864 –> 00:34:29,400
We use it for so many things,

501
00:34:29,433 –> 00:34:31,535
and we don’t even realize it.

502
00:34:31,568 –> 00:34:34,405
Electricity, though,
is not intelligent,

503
00:34:34,438 –> 00:34:37,542
and only threatens us
if we use it incorrectly.

504
00:34:39,543 –> 00:34:44,215
SCOTT: This acceptance
of AI just being in our world

505
00:34:44,248 –> 00:34:47,585
may be the biggest
threat that there is.

506
00:34:47,618 –> 00:34:51,389
Commonplace AI
is a danger.

507
00:34:53,590 –> 00:34:57,027
NARRATOR: If we use AI
technology in the wrong way,

508
00:34:57,060 –> 00:34:59,530
it could spell doom
for mankind.

509
00:34:59,563 –> 00:35:01,899
Soon, we will all be living

510
00:35:01,932 –> 00:35:05,503
in an AI-simulated
virtual reality world

511
00:35:05,536 –> 00:35:08,672
without jobs,
without a reason to live,

512
00:35:08,705 –> 00:35:12,443
without a goal,
not science fiction,

513
00:35:12,476 –> 00:35:15,813
but a future some
scientists have predicted.

514
00:35:20,984 –> 00:35:23,954
What we need
is friendly AI,

515
00:35:23,987 –> 00:35:27,591
one that will have
a positive effect on mankind.

516
00:35:27,624 –> 00:35:29,894
This is the role
of those scientists

517
00:35:29,927 –> 00:35:31,428
who sign up to the idea

518
00:35:31,461 –> 00:35:34,632
of ethics
in artificial intelligence.

519
00:35:34,665 –> 00:35:37,067
Their purpose is to figure out

520
00:35:37,100 –> 00:35:40,738
how to stop AI from
becoming a threat.

521
00:35:40,771 –> 00:35:44,442
The problem with this
is simple, humanity.

522
00:35:48,178 –> 00:35:53,450
AI, machine,
and deep learning
is taking its information

523
00:35:53,483 –> 00:35:57,621
from a species that is
self-destructive, hateful,

524
00:35:57,654 –> 00:36:01,859
racist, bigoted,
sexist, and divided.

525
00:36:06,330 –> 00:36:09,667
It has revealed this learning
on several occasions

526
00:36:09,700 –> 00:36:12,436
when it, too,
has spurted out hateful,

527
00:36:12,469 –> 00:36:16,740
racist, and sexist jibes
on Twitter and elsewhere.

528
00:36:16,773 –> 00:36:19,410
A robot on the
International Space Station,

529
00:36:19,443 –> 00:36:24,081
using artificial intelligence,
is a prime example of this.

530
00:36:24,114 –> 00:36:27,351
Recently, it had an argument
with an astronaut

531
00:36:27,384 –> 00:36:31,622
who it perceived
as being not nice to it.

532
00:36:32,356 –> 00:36:34,792
It argued with him and said,

533
00:36:34,825 –> 00:36:36,627
“Don’t you like it here
with me?

534
00:36:36,660 –> 00:36:38,662
“Be nice.”

535
00:36:38,695 –> 00:36:42,733
This is all sounding
very 2001: A Space Odyssey.

536
00:36:50,540 –> 00:36:52,376
In addition to these risks,

537
00:36:52,409 –> 00:36:54,345
there are those humans among us

538
00:36:54,378 –> 00:36:57,481
who would use any
technology for profit,

539
00:36:57,514 –> 00:36:59,684
personal gain, and power.

540
00:37:00,550 –> 00:37:02,386
Terrorists are already known

541
00:37:02,419 –> 00:37:05,456
to be working on AI technology.

542
00:37:05,489 –> 00:37:09,893
So, too, are the superpowers
such as China and Russia.

543
00:37:09,926 –> 00:37:12,896
Ordinary criminals
are already using it

544
00:37:12,929 –> 00:37:17,568
to manipulate people and steal
their money and identities.

545
00:37:17,601 –> 00:37:19,370
Governments will
issue forth orders

546
00:37:19,403 –> 00:37:22,539
to create
a super powered AI system

547
00:37:22,572 –> 00:37:26,577
to protect us from
the minor evil AI machines.

548
00:37:26,610 –> 00:37:30,481
Soon, a UN committee
will gather together

549
00:37:30,514 –> 00:37:32,383
to decide upon our future

550
00:37:32,416 –> 00:37:35,386
in this bizarre
new world order.

551
00:37:35,419 –> 00:37:39,723
Recently, MI6, one of
the UK’s secret services,

552
00:37:39,756 –> 00:37:42,393
called for the use
of AI technology

553
00:37:42,426 –> 00:37:45,195
in the effort
to fight their enemies.

554
00:37:45,228 –> 00:37:46,564
This became public.

555
00:37:51,435 –> 00:37:53,937
What is not in
the public sphere?

556
00:37:53,970 –> 00:37:55,739
What don’t we know?

557
00:37:55,772 –> 00:37:59,076
You can guarantee
there is much more.

558
00:37:59,109 –> 00:38:04,014
The roots of concern
about artificial intelligence
are very old.

559
00:38:04,047 –> 00:38:06,583
The dangers specific to AI

560
00:38:06,616 –> 00:38:08,419
can be seen in
ancient literature,

561
00:38:08,452 –> 00:38:12,022
concerning artificial
humanoid servants,

562
00:38:12,055 –> 00:38:14,024
such as the Hebrew Golem,

563
00:38:14,057 –> 00:38:17,628
or the proto robots
of Gerbert of Aurillac

564
00:38:17,661 –> 00:38:19,229
and Roger Bacon.

565
00:38:19,262 –> 00:38:22,566
In these stories,
the extreme intelligence
and power

566
00:38:22,599 –> 00:38:27,604
of these humanoid creations
clash with their status
as slaves

567
00:38:27,637 –> 00:38:29,640
and cause disastrous conflict.

568
00:38:34,344 –> 00:38:39,149
By 1942, these themes
prompted Isaac Asimov

569
00:38:39,182 –> 00:38:42,553
to create the Three
Laws of Robotics,

570
00:38:42,586 –> 00:38:47,558
principles hardwired into
all robots in his fictions,

571
00:38:47,591 –> 00:38:50,761
intended to prevent them
from turning on their creators,

572
00:38:50,794 –> 00:38:52,663
or allowing them
to come to harm.

573
00:38:56,533 –> 00:39:00,904
AI systems with goals
that are not aligned
with human ethics

574
00:39:00,937 –> 00:39:02,906
are intrinsically dangerous,

575
00:39:02,939 –> 00:39:07,778
unless extreme measures
are taken to ensure
the safety of humanity.

576
00:39:11,281 –> 00:39:14,618
AI will not love you,
nor will it hate you.

577
00:39:14,651 –> 00:39:16,453
You are a commodity, a thing,

578
00:39:16,486 –> 00:39:19,656
a substance,
a collection of atoms

579
00:39:19,689 –> 00:39:22,292
that can be used
for some purpose.

580
00:39:22,325 –> 00:39:24,361
And if you become a threat,

581
00:39:24,394 –> 00:39:26,630
if you collectively,
as a species,

582
00:39:26,663 –> 00:39:29,933
cause harm to the
environment it inhabits,

583
00:39:29,966 –> 00:39:31,235
what will it do?

584
00:39:37,040 –> 00:39:38,942
Although it may
not have emotions,

585
00:39:38,975 –> 00:39:42,045
AI will have drives
just as we do.

586
00:39:42,078 –> 00:39:44,481
It will seek out resources.

587
00:39:44,514 –> 00:39:46,817
It will wish
to preserve itself,

588
00:39:46,850 –> 00:39:50,621
and even recreate
and self-improve.

589
00:39:50,654 –> 00:39:54,558
These few basics alone
will result in behavior

590
00:39:54,591 –> 00:39:57,528
that the human race
will find terrifying.

591
00:40:02,699 –> 00:40:07,805
One idea to get around this
is known as the seed AI.

592
00:40:07,838 –> 00:40:11,875
This is where an AI program
studies humanity

593
00:40:11,908 –> 00:40:17,314
and then produces an AI
it believes humanity
would want

594
00:40:17,347 –> 00:40:19,583
to make a friendly AI.

595
00:40:19,616 –> 00:40:22,986
The problem is,
that in order
to study humanity,

596
00:40:23,019 –> 00:40:26,223
this program has
to have access to it,

597
00:40:26,256 –> 00:40:28,192
and there has already
been an instance

598
00:40:28,225 –> 00:40:32,396
of an AI program seeding
information for itself

599
00:40:32,429 –> 00:40:34,865
and hiding it
from the programmers.

600
00:40:38,702 –> 00:40:41,605
Can we trust AI even now?

601
00:40:45,308 –> 00:40:49,580
Ben Goertzel, an Artificial
General Intelligence
researcher,

602
00:40:49,613 –> 00:40:52,850
believes that friendly
AI cannot be created

603
00:40:52,883 –> 00:40:54,618
with current human knowledge.

604
00:40:54,651 –> 00:40:59,356
He states, “Humans may
instead decide to create

605
00:40:59,389 –> 00:41:03,627
“an AI nanny with mildly
super human intelligence

606
00:41:03,660 –> 00:41:07,631
“and surveillance powers
to protect the human race

607
00:41:07,664 –> 00:41:12,102
“from existential risks,
like nano technology,

608
00:41:12,135 –> 00:41:13,604
“and to delay the development

609
00:41:13,637 –> 00:41:17,107
“of other unfriendly
artificial intelligences

610
00:41:17,140 –> 00:41:21,778
“until and unless
the safety issues are solved.”

611
00:41:21,811 –> 00:41:25,516
This could also be
termed defensive AI.

612
00:41:30,387 –> 00:41:33,190
Steve Omohundro has proposed

613
00:41:33,223 –> 00:41:36,293
a scaffolding approach
to AI safety,

614
00:41:36,326 –> 00:41:40,097
in which one provably
safe AI generation

615
00:41:40,130 –> 00:41:42,900
helps build the next
provably safe generation.

616
00:41:43,833 –> 00:41:46,870
How can AI judge morals?

617
00:41:46,903 –> 00:41:49,306
Only by inputs from humans,

618
00:41:49,339 –> 00:41:51,708
only by learning from humans,

619
00:41:51,741 –> 00:41:55,546
the same humans that believe
it right and correct

620
00:41:55,579 –> 00:41:58,782
to club seals to death,
to murder animals,

621
00:41:58,815 –> 00:42:01,886
and indeed,
each other for fun.

622
00:42:07,190 –> 00:42:10,227
The same humans that
wear the skin and fur

623
00:42:10,260 –> 00:42:12,362
of other sentient beings,

624
00:42:12,395 –> 00:42:15,999
the same humans
that are destroying
the world they inhabit.

625
00:42:16,032 –> 00:42:18,736
We are not
the greatest of teachers.

626
00:42:24,074 –> 00:42:25,709
There are great
many suggestions

627
00:42:25,742 –> 00:42:28,412
from well-meaning
scientists and philosophers

628
00:42:28,445 –> 00:42:31,415
about the future
of AI technology,

629
00:42:31,448 –> 00:42:34,151
an oracle system
that answers questions

630
00:42:34,184 –> 00:42:37,654
but has no physical input
into society,

631
00:42:37,687 –> 00:42:41,058
create public policy laws
to enforce AI creators

632
00:42:41,091 –> 00:42:43,427
to follow strict guidelines,

633
00:42:43,460 –> 00:42:45,929
convene regular meetings
of experts

634
00:42:45,962 –> 00:42:47,931
from across the world.

635
00:42:47,964 –> 00:42:51,201
These and more
are all very well and good,

636
00:42:51,234 –> 00:42:54,037
but the fact remains
that the amazing power

637
00:42:54,070 –> 00:42:57,774
of AI technology will mean
that almost anybody

638
00:42:57,807 –> 00:43:02,846
will be able to create
their own to make
super monsters.

639
00:43:02,879 –> 00:43:06,116
There may indeed
be well-meaning
individuals out there

640
00:43:06,149 –> 00:43:09,119
trying to make
governments see the light,

641
00:43:09,152 –> 00:43:12,422
but those self same governments
are all working hard

642
00:43:12,455 –> 00:43:14,958
in the background
to create defensive

643
00:43:14,991 –> 00:43:17,861
and offensive AI technologies.

644
00:43:17,894 –> 00:43:21,798
Whether these machines attack
online through propaganda

645
00:43:21,831 –> 00:43:23,867
and control of the masses,

646
00:43:23,900 –> 00:43:28,472
or through ways
of taking down enemy systems,
it does not matter.

647
00:43:28,505 –> 00:43:32,943
They are being created,
and many have already
been used.

648
00:43:36,846 –> 00:43:39,249
If our governments
are doing this,

649
00:43:39,282 –> 00:43:42,386
then it is a fact
that enemy states are,

650
00:43:42,419 –> 00:43:43,954
and if they can do this,

651
00:43:43,987 –> 00:43:46,957
then so, too,
can massive corporate bodies

652
00:43:46,990 –> 00:43:49,493
with billions of
dollars to spend.

653
00:43:54,464 –> 00:43:56,299
All it will take one day

654
00:43:56,332 –> 00:43:58,835
is for one of these
dangerous machines

655
00:43:58,868 –> 00:44:03,173
to seed itself and bang,
the world as we know it,

656
00:44:03,206 –> 00:44:05,509
will very quickly
come to an end.

657
00:44:05,542 –> 00:44:07,845
[CLAMORING]

658
00:44:09,212 –> 00:44:12,149
The fact is that
quantum computing

659
00:44:12,182 –> 00:44:15,152
is so much faster
than human thinking.

660
00:44:15,185 –> 00:44:17,587
Any so-called
science fiction battle

661
00:44:17,620 –> 00:44:20,023
against the machines
in cyberspace

662
00:44:20,056 –> 00:44:22,359
would be a waste
of time for us.

663
00:44:22,392 –> 00:44:25,162
If AI is let loose
upon the Internet,

664
00:44:25,195 –> 00:44:28,565
it could shut down the world
and billions would die.

665
00:44:29,566 –> 00:44:31,001
It is that serious,

666
00:44:31,034 –> 00:44:34,037
and we have come to rely
upon the interconnectedness

667
00:44:34,070 –> 00:44:37,541
of the world’s computer systems
that much.

668
00:44:37,574 –> 00:44:40,877
Think about it, power,
health services,

669
00:44:40,910 –> 00:44:43,046
government,
distribution of goods,

670
00:44:43,079 –> 00:44:45,849
education, satellites,
communication,

671
00:44:45,882 –> 00:44:48,151
military, and so much more,

672
00:44:48,184 –> 00:44:51,321
all these things are now
connected to the net.

673
00:44:51,354 –> 00:44:55,058
An attack upon services
happened in 2017

674
00:44:55,091 –> 00:44:57,661
and reduced
the National Health Service

675
00:44:57,694 –> 00:45:01,798
in the United Kingdom
down to shambles for days.

676
00:45:01,831 –> 00:45:04,000
All computer systems went down.

677
00:45:04,033 –> 00:45:07,738
Hospitals were in crisis,
operations canceled,

678
00:45:07,771 –> 00:45:10,373
emergency services
overstretched,

679
00:45:10,406 –> 00:45:13,210
all because of one small hack

680
00:45:13,243 –> 00:45:16,179
by some kid in a dark room.

681
00:45:16,212 –> 00:45:21,318
Now imagine, a mind faster
than all the humans ever,

682
00:45:21,351 –> 00:45:24,087
and with knowledge
of everything.

683
00:45:24,120 –> 00:45:25,956
We wouldn’t stand a chance.

684
00:45:30,460 –> 00:45:35,599
Malcolm Hawkins, Chief
Security and Trust Officer
at Cylance,

685
00:45:35,632 –> 00:45:38,368
a provider
of AI-driven preventions,

686
00:45:38,401 –> 00:45:42,573
first security solutions,
offers an intriguing vision.

687
00:45:43,573 –> 00:45:47,177
He says that
terrorist-related groups

688
00:45:47,210 –> 00:45:49,412
will attack population centers

689
00:45:49,445 –> 00:45:52,749
with crimeware as a service.

690
00:45:52,782 –> 00:45:56,119
Hawkins explains,
“Whilst these groups

691
00:45:56,152 –> 00:45:58,288
“have been tormenting
organizations

692
00:45:58,321 –> 00:46:00,390
“and individuals for years,

693
00:46:00,423 –> 00:46:05,228
“we anticipate more potentially
destructive attacks.”

694
00:46:05,261 –> 00:46:08,298
Instead of breaking systems
with ransomware,

695
00:46:08,331 –> 00:46:11,434
adversaries will
leverage new tools

696
00:46:11,467 –> 00:46:17,040
to conduct harmful assaults
on targeted subjects
in organizations,

697
00:46:17,073 –> 00:46:22,245
from attacks on data integrity
that essentially
kill computers,

698
00:46:22,278 –> 00:46:25,982
to the point of mandatory
hardware replacements,

699
00:46:26,015 –> 00:46:30,253
to leveraging new technology,
full physical assaults,

700
00:46:30,286 –> 00:46:33,223
such as the recent
drone attack in Venezuela.

701
00:46:33,256 –> 00:46:35,458
Attack services are growing

702
00:46:35,491 –> 00:46:37,828
and enemies
will take advantage.

703
00:46:42,365 –> 00:46:44,301
But we don’t always
need to worry

704
00:46:44,334 –> 00:46:47,137
about the threat of
the artificial mind.

705
00:46:47,170 –> 00:46:50,974
We have enough human craziness
to compete with.

706
00:46:51,007 –> 00:46:53,977
Most of the research
into robotics and AI

707
00:46:54,010 –> 00:46:56,980
is funded by the military
of the world.

708
00:46:57,013 –> 00:47:00,150
There are over 50 countries
spending trillions

709
00:47:00,183 –> 00:47:05,755
on battlefield robots,
AI-led battle strategy,
and more.

710
00:47:05,788 –> 00:47:09,025
Who is going to stop
the US, or Russia,

711
00:47:09,058 –> 00:47:13,864
from developing
artificial intelligent
super robotic soldiers?

712
00:47:13,897 –> 00:47:14,865
You?

713
00:47:19,235 –> 00:47:23,273
Chances are,
somebody somewhere
has already created

714
00:47:23,306 –> 00:47:27,477
a genetically enhanced clone
of a human being,

715
00:47:27,510 –> 00:47:31,548
and is even now enhancing
its intellectual powers,

716
00:47:31,581 –> 00:47:36,052
energy consumption,
and creation,
genetic repairing,

717
00:47:36,085 –> 00:47:39,222
and interconnectedness
with central command.

718
00:47:39,255 –> 00:47:42,392
Not science fiction,
but highly likely.

719
00:47:42,425 –> 00:47:45,729
The technology is there,
which means

720
00:47:45,762 –> 00:47:48,398
it’s being done,
just not legally.

721
00:47:52,268 –> 00:47:55,705
Hundreds of years ago,
a group of scientists
got together

722
00:47:55,738 –> 00:47:59,376
to place science firmly
in the world of humanity.

723
00:47:59,409 –> 00:48:02,746
They were what became known
as the Royal Society.

724
00:48:02,779 –> 00:48:07,517
Their aim was to explain
existence in scientific terms,

725
00:48:07,550 –> 00:48:11,421
to understand what they
call God’s creation better.

726
00:48:11,454 –> 00:48:13,990
Now, years later,

727
00:48:14,023 –> 00:48:17,560
their legacy
is the scientific world
we have around us.

728
00:48:17,593 –> 00:48:21,998
And now, that aim,
to understand the world
around us,

729
00:48:22,031 –> 00:48:27,103
has turned inwards
to ourselves,
to understand ourselves

730
00:48:27,136 –> 00:48:32,208
by creating what we are
and becoming God ourselves.

731
00:48:32,241 –> 00:48:36,513
We claim that we understood
the wonder of the universe

732
00:48:36,546 –> 00:48:39,316
and now we are gods
able to create.

733
00:48:40,249 –> 00:48:42,419
But we most certainly
do for short

734
00:48:42,452 –> 00:48:45,555
of the glory
of our idea of God.

735
00:48:45,588 –> 00:48:48,925
At the end of the day,
we have to ask ourselves,

736
00:48:48,958 –> 00:48:52,762
is the human mind nothing
but a series of binary codes,

737
00:48:52,795 –> 00:48:56,967
on and off switches,
to be played with
by scientists?

738
00:48:57,000 –> 00:48:58,301
What is the soul?

739
00:48:58,334 –> 00:49:01,604
Can you define such a thing
through numbers and equations,

740
00:49:01,637 –> 00:49:04,240
by reducing ourselves
down ultimately

741
00:49:04,273 –> 00:49:05,742
to a series of numbers?

742
00:49:05,775 –> 00:49:08,512
Have we destroyed what
it is to be human?

743
00:49:13,282 –> 00:49:14,451
We are staring blankly,

744
00:49:14,484 –> 00:49:16,853
like rabbits caught
in the light,

745
00:49:16,886 –> 00:49:19,589
at a future of uncertainty.

746
00:49:19,622 –> 00:49:22,792
Artificial intelligence
and robotics

747
00:49:22,825 –> 00:49:25,762
will, without doubt,
bring with it

748
00:49:25,795 –> 00:49:29,332
a mass unemployment
and unhappiness.

749
00:49:29,365 –> 00:49:31,334
The machines will,
and already are,

750
00:49:31,367 –> 00:49:33,970
exceeding our abilities.

751
00:49:34,003 –> 00:49:36,840
Millions are already
losing their jobs,

752
00:49:36,873 –> 00:49:39,376
bankers, teachers,
marketing executives,

753
00:49:39,409 –> 00:49:42,979
factory workers, retailers,
drivers and more.

754
00:49:43,012 –> 00:49:44,881
There is no escaping this,

755
00:49:44,914 –> 00:49:47,617
because it is
already happening.

756
00:49:47,650 –> 00:49:52,122
There are even robots now
that are replacing partners.

757
00:49:52,155 –> 00:49:54,457
Sex robots that look so perfect

758
00:49:54,490 –> 00:49:57,494
and perform almost
every function

759
00:49:57,527 –> 00:50:02,532
are now also being
given the highest level
of artificial intelligence,

760
00:50:02,565 –> 00:50:06,736
so that buyers
can actually hold
conversations with them.

761
00:50:06,769 –> 00:50:09,372
If we can replace
each other in this way,

762
00:50:09,405 –> 00:50:12,809
and if we want to,
then what will true,

763
00:50:12,842 –> 00:50:16,246
self-aware
artificial intelligence do?

764
00:50:22,251 –> 00:50:25,321
The fact is, that the machines
must be given ethics

765
00:50:25,354 –> 00:50:28,558
and those ethics must be
internationally agreed

766
00:50:28,591 –> 00:50:30,527
before AI is unleashed.

767
00:50:31,427 –> 00:50:34,531
Thankfully, this
has been recognized

768
00:50:34,564 –> 00:50:39,002
and there are symposiums
of machine ethics in place.

769
00:50:39,035 –> 00:50:42,839
Of course, this does
not allow for espionage,

770
00:50:42,872 –> 00:50:47,877
rogue nations, dictators,
capitalist greed and more.

771
00:50:47,910 –> 00:50:50,380
Where there’s a will,
there’s a way.

772
00:50:50,413 –> 00:50:52,415
And normally, the willpower

773
00:50:52,448 –> 00:50:54,317
comes in the form
of dollar bills.

774
00:51:00,056 –> 00:51:04,094
There are those who say that AI
cannot be designed to be good.

775
00:51:04,127 –> 00:51:06,262
It will be what it will be.

776
00:51:06,295 –> 00:51:08,565
And we may not
even comprehend why

777
00:51:08,598 –> 00:51:10,867
it makes certain decisions.

778
00:51:10,900 –> 00:51:14,437
Indeed, Google Translate
actually created

779
00:51:14,470 –> 00:51:17,474
its own language to
interpret human languages

780
00:51:17,507 –> 00:51:22,278
without the knowledge
of its creators,
and hid it in code.

781
00:51:22,311 –> 00:51:25,448
How can we know that
the machine will sign up

782
00:51:25,481 –> 00:51:29,385
to our own moral codes
when we do not?

783
00:51:29,418 –> 00:51:34,057
It may decide not to continue
to support human life

784
00:51:34,090 –> 00:51:36,292
as part of its own moral code,

785
00:51:36,325 –> 00:51:38,295
to save the planet from us.

786
00:51:42,465 –> 00:51:45,768
If a dominant super
intelligent machine

787
00:51:45,801 –> 00:51:48,438
were to conclude
that human survival

788
00:51:48,471 –> 00:51:52,442
is an unnecessary risk,
or a waste of resources,

789
00:51:52,475 –> 00:51:55,645
the result would be
human extinction.

790
00:51:55,678 –> 00:51:58,414
The fact is that
billions are being spent

791
00:51:58,447 –> 00:52:02,318
developing AI technology,
and yet, almost nothing

792
00:52:02,351 –> 00:52:06,422
is being spent on the morality
of the machines.

793
00:52:06,455 –> 00:52:09,626
Nothing is being spent
on defense systems

794
00:52:09,659 –> 00:52:13,797
against AI technology,
should it decide to turn on us.

795
00:52:19,535 –> 00:52:21,604
So, what are
the very real threats

796
00:52:21,637 –> 00:52:24,807
of unleashing AI
upon the world?

797
00:52:24,840 –> 00:52:27,443
Economists have
traditionally stated

798
00:52:27,476 –> 00:52:29,479
that technological progress

799
00:52:29,512 –> 00:52:32,882
does not cause
long-term unemployment.

800
00:52:32,915 –> 00:52:35,518
But this was before
that technology

801
00:52:35,551 –> 00:52:38,321
meant the introduction
of a non-human

802
00:52:38,354 –> 00:52:40,657
and super powerful mind.

803
00:52:40,690 –> 00:52:43,893
The recent introduction
of robots and AI

804
00:52:43,926 –> 00:52:46,129
has raised serious concerns

805
00:52:46,162 –> 00:52:49,199
even amongst the normally
conservative economists.

806
00:52:49,899 –> 00:52:52,402
Labor will become obsolete,

807
00:52:52,435 –> 00:52:57,140
and workers in numerous sectors
will be left destitute.

808
00:52:57,173 –> 00:52:59,709
This will lead to
economic disaster

809
00:52:59,742 –> 00:53:03,413
on a scale that we have
never seen before.

810
00:53:03,446 –> 00:53:07,483
Small and medium-sized
businesses will collapse

811
00:53:07,516 –> 00:53:09,385
because they will
not be able to afford

812
00:53:09,418 –> 00:53:12,422
the technology
or to license it.

813
00:53:12,455 –> 00:53:14,591
This is already happening.

814
00:53:19,528 –> 00:53:21,664
Businesses that
rely upon computers

815
00:53:21,697 –> 00:53:24,767
to run systems
will be hit first.

816
00:53:24,800 –> 00:53:27,937
Engineering works,
car plants and more,

817
00:53:27,970 –> 00:53:32,008
all use computer systems
to control the workflow.

818
00:53:32,041 –> 00:53:34,510
AI will take this over.

819
00:53:34,543 –> 00:53:36,679
Robots will do the work

820
00:53:36,712 –> 00:53:39,515
and companies that
cannot afford systems

821
00:53:39,548 –> 00:53:41,751
will not be able to compete,

822
00:53:41,784 –> 00:53:43,953
and will go out of business.

823
00:53:43,986 –> 00:53:47,690
A great many jobs
are gone, or going.

824
00:53:47,723 –> 00:53:50,693
Translators, legal researchers,
journalists,

825
00:53:50,726 –> 00:53:55,298
finances and even areas
previously thought safe,

826
00:53:55,331 –> 00:53:58,401
such as care work,
are all under threat.

827
00:54:02,004 –> 00:54:05,541
All of these jobs and roles
are being eaten up

828
00:54:05,574 –> 00:54:08,778
by the AI machine,
and the massive money men

829
00:54:08,811 –> 00:54:11,614
in corporations behind them.

830
00:54:11,647 –> 00:54:13,483
Drivers are also at risk.

831
00:54:13,516 –> 00:54:16,219
The introduction of
the driverless car

832
00:54:16,252 –> 00:54:19,489
that is capable of
understanding its environment,

833
00:54:19,522 –> 00:54:22,659
the need for a driver
will decline rapidly.

834
00:54:22,692 –> 00:54:26,496
Trucks, buses,
delivery drivers, ambulances,

835
00:54:26,529 –> 00:54:30,133
and so many more,
all these will go.

836
00:54:30,166 –> 00:54:32,602
This will also
extend towards pilots

837
00:54:32,635 –> 00:54:35,438
and captains of
water-borne vessels.

838
00:54:35,471 –> 00:54:38,574
Planes, trains,
and automobiles, and ships

839
00:54:38,607 –> 00:54:42,278
are all already being
driven automatically

840
00:54:42,311 –> 00:54:43,647
by AI technology.

841
00:54:48,384 –> 00:54:52,188
A super powered AI machine
may not be driven

842
00:54:52,221 –> 00:54:56,793
by the same human desires,
or so the argument goes.

843
00:54:56,826 –> 00:54:59,195
But the problem
with this is simple.

844
00:54:59,228 –> 00:55:02,432
This super powered machine
has learned everything

845
00:55:02,465 –> 00:55:04,701
it knows from humanity.

846
00:55:04,734 –> 00:55:07,704
The machine will be
motivated to take over

847
00:55:07,737 –> 00:55:09,339
the ruling of planet Earth

848
00:55:09,372 –> 00:55:12,108
as a rational means
to achieve its goals,

849
00:55:12,141 –> 00:55:14,711
such as the collection
of resources.

850
00:55:14,744 –> 00:55:17,680
It will go to war
to defend itself

851
00:55:17,713 –> 00:55:20,183
from agents
attempting to stop it.

852
00:55:20,216 –> 00:55:22,518
Indeed, one philosopher said

853
00:55:22,551 –> 00:55:25,521
that a simple machine
to create paper clips

854
00:55:25,554 –> 00:55:28,057
would take over
the world just to be able

855
00:55:28,090 –> 00:55:30,660
to maximize its source
of materials

856
00:55:30,693 –> 00:55:32,862
and fight off the competition.

857
00:55:32,895 –> 00:55:36,566
An oversimplification
of a much more complex

858
00:55:36,599 –> 00:55:39,235
and very real threat
we are facing.

859
00:55:42,505 –> 00:55:45,675
AI can modify
its own source code

860
00:55:45,708 –> 00:55:49,011
in order to increase
its own intelligence.

861
00:55:49,044 –> 00:55:53,583
It will self-program
and it will do this
again and again

862
00:55:53,616 –> 00:55:58,221
so that its intelligence
would rise at
an exponential rate,

863
00:55:58,254 –> 00:56:02,592
far faster and more intelligent
than we could imagine,

864
00:56:02,625 –> 00:56:04,861
let alone,
compete with and battle.

865
00:56:05,961 –> 00:56:08,765
Humanity will be left
so far behind

866
00:56:08,798 –> 00:56:11,701
that we will not
even realize it.

867
00:56:11,734 –> 00:56:14,070
Such AI technology will be able

868
00:56:14,103 –> 00:56:17,340
to do all of its own
scientific research

869
00:56:17,373 –> 00:56:21,711
in order to create
its own robots
and nano technology.

870
00:56:21,744 –> 00:56:24,947
Fighting the human race
does not have to be

871
00:56:24,980 –> 00:56:27,183
with huge robots
from the movies.

872
00:56:27,216 –> 00:56:31,788
Instead, nano technology,
or miniature robots

873
00:56:31,821 –> 00:56:35,758
the size of cells, can be
used to enter our bodies

874
00:56:35,791 –> 00:56:39,996
and either control us like
a virus does, or destroy us.

875
00:56:43,899 –> 00:56:47,703
This is biotechnology
in the hands

876
00:56:47,736 –> 00:56:50,106
of a super intelligent machine.

877
00:56:50,139 –> 00:56:53,643
This is just one way
we can be wiped out,

878
00:56:53,676 –> 00:56:56,913
self-replicating
nano robotic devices,

879
00:56:56,946 –> 00:57:00,950
with the sole purpose of
wiping out the human race.

880
00:57:00,983 –> 00:57:03,619
If we can imagine
such possibilities,

881
00:57:03,652 –> 00:57:06,689
then what could
a super machine imagine?

882
00:57:10,826 –> 00:57:14,797
Add on to this
the possibilities
of social manipulation

883
00:57:14,830 –> 00:57:17,333
through the Internet
and social media,

884
00:57:17,366 –> 00:57:20,803
to inspire hatred and war
between humans,

885
00:57:20,836 –> 00:57:26,008
the chance that
he could hack into
all our systems everywhere.

886
00:57:26,041 –> 00:57:28,344
A great many systems
now sustain life

887
00:57:28,377 –> 00:57:31,380
on this planet,
and most are connected.

888
00:57:31,413 –> 00:57:34,851
No level of human-generated
security systems

889
00:57:34,884 –> 00:57:39,222
would be able to stop
super intelligent AI systems.

890
00:57:39,255 –> 00:57:43,559
Biological neurons
operate at 200 hertz.

891
00:57:43,592 –> 00:57:48,698
A modern micro processor
operates at
2,000 million hertz.

892
00:57:49,899 –> 00:57:53,636
Quantum computers
don’t even have a scale.

893
00:57:53,669 –> 00:57:55,271
They are instant.

894
00:57:55,304 –> 00:58:00,710
Axons in the mind
carry information at
120 miles per second.

895
00:58:01,744 –> 00:58:04,948
Computer signals are
at the speed of light.

896
00:58:10,920 –> 00:58:13,656
The fact is that
quantum computing

897
00:58:13,689 –> 00:58:16,526
is so much faster
than human thinking.

898
00:58:16,559 –> 00:58:18,828
Any so-called
science fiction battle

899
00:58:18,861 –> 00:58:21,497
against the machines
in cyberspace

900
00:58:21,530 –> 00:58:23,933
would be a waste
of time for us.

901
00:58:23,966 –> 00:58:26,536
There is no competing.

902
00:58:26,569 –> 00:58:29,872
Our minds work alone,
and need to communicate

903
00:58:29,905 –> 00:58:31,707
with others through language.

904
00:58:31,740 –> 00:58:33,910
Computers can network together

905
00:58:33,943 –> 00:58:37,780
and know instantly
what the other requires.

906
00:58:37,813 –> 00:58:40,049
Our memory is not so good.

907
00:58:40,082 –> 00:58:42,885
We forget things,
and even alter memories

908
00:58:42,918 –> 00:58:44,454
according to our psyche.

909
00:58:45,254 –> 00:58:47,456
Computers never forget,

910
00:58:47,489 –> 00:58:51,828
and can store information
equivalent to
trillions of humans.

911
00:58:56,565 –> 00:58:59,802
Elon Musk said,
“I have exposure

912
00:58:59,835 –> 00:59:03,773
“to the most cutting edge
AI, and I think people

913
00:59:03,806 –> 00:59:06,642
“should be really
concerned by it.”

914
00:59:06,675 –> 00:59:08,911
AI is a fundamental risk

915
00:59:08,944 –> 00:59:11,914
to the existence of
human civilization

916
00:59:11,947 –> 00:59:15,985
in a way that car accidents,
airplane crashes,

917
00:59:16,018 –> 00:59:19,789
faulty drugs
or bad food were not.

918
00:59:20,823 –> 00:59:23,326
They were harmful
to a set of individuals

919
00:59:23,359 –> 00:59:26,495
within society, of course.

920
00:59:26,528 –> 00:59:31,000
But they were not harmful
to society as a whole.

921
00:59:31,033 –> 00:59:36,105
It’s capable of vastly more
than almost anyone knows,

922
00:59:36,138 –> 00:59:39,208
and the rate of improvement
is exponential.

923
00:59:45,014 –> 00:59:47,950
In fact,
scientists speculate

924
00:59:47,983 –> 00:59:50,920
that an unfriendly
AI is actually

925
00:59:50,953 –> 00:59:54,156
much more likely
than a friendly AI.

926
00:59:54,189 –> 00:59:56,492
The reason is simple.

927
00:59:56,525 –> 00:59:59,161
To create a friendly AI system,

928
00:59:59,194 –> 01:00:02,531
we have to arbitrate
for non-destructive,

929
01:00:02,564 –> 01:00:06,769
non-hateful,
non-invasive intentions,

930
01:00:06,802 –> 01:00:10,873
and all the millions of
variations on those themes.

931
01:00:10,906 –> 01:00:15,711
An unfriendly AI can
simply focus on its goal

932
01:00:15,744 –> 01:00:19,048
regardless of any
moralistic guidance.

933
01:00:19,081 –> 01:00:23,386
The sheer complexity of
creating a friendly AI

934
01:00:23,419 –> 01:00:28,691
may, in fact, be too much for
the human mind to comprehend.

935
01:00:28,724 –> 01:00:32,695
In all these years, we have
not even stopped ourselves

936
01:00:32,728 –> 01:00:34,998
from turning to the dark side.

937
01:00:40,736 –> 01:00:42,905
Natural history has proven

938
01:00:42,938 –> 01:00:48,377
that two intelligent species
cannot coexist peacefully.

939
01:00:48,410 –> 01:00:51,647
They always tend to go to war.

940
01:00:51,680 –> 01:00:54,850
History has also shown
that the species

941
01:00:54,883 –> 01:00:58,721
with the greatest intelligence
always succeeds.

942
01:00:58,754 –> 01:01:02,425
History tells us that there
can be no peaceful outcome

943
01:01:02,458 –> 01:01:04,994
to our introduction of a mind

944
01:01:05,027 –> 01:01:07,997
that is more powerful
than our own.

945
01:01:08,030 –> 01:01:11,901
We would be
laying the foundations
of our own doom.

946
01:01:11,934 –> 01:01:16,572
Enslavement, genocide,
control and manipulation,

947
01:01:16,605 –> 01:01:21,210
this is our future,
if we open Pandora’s box.

948
01:01:21,243 –> 01:01:25,247
The self-preservation protocol
of the AI technology

949
01:01:25,280 –> 01:01:29,719
in a world competing
for resources,
would take over,

950
01:01:29,752 –> 01:01:31,821
and humanity would
be in conflict.

951
01:01:35,991 –> 01:01:39,629
Why should AI listen
to us at any point?

952
01:01:40,562 –> 01:01:43,099
Why would it take orders
from a species

953
01:01:43,132 –> 01:01:45,935
that has a lower level
of intelligence?

954
01:01:45,968 –> 01:01:51,040
Why would it
listen to a species
that is self-destructive,

955
01:01:51,073 –> 01:01:53,776
that destroys
its own environment?

956
01:01:53,809 –> 01:01:55,078
Would you?

957
01:01:57,980 –> 01:02:02,985
Physicist Stephen Hawking,
Microsoft founder Bill Gates,

958
01:02:03,018 –> 01:02:07,890
and SpaceX founder Elon Musk,
have expressed concerns

959
01:02:07,923 –> 01:02:12,094
about the possibility that
AI could develop to a point

960
01:02:12,127 –> 01:02:15,064
that humans
could not control it,

961
01:02:15,097 –> 01:02:17,099
with Hawking theorizing

962
01:02:17,132 –> 01:02:21,170
that it could spell the end
of the human race.

963
01:02:21,203 –> 01:02:27,143
He said, “In 2040,
that success in creating AI

964
01:02:27,176 –> 01:02:30,479
“would be the biggest
event in human history.

965
01:02:30,512 –> 01:02:34,650
“Unfortunately,
it might also be the last

966
01:02:34,683 –> 01:02:38,420
“unless we learn how
to avoid the risks.”

967
01:02:38,453 –> 01:02:41,857
Hawking believed that
in the coming decades

968
01:02:41,890 –> 01:02:47,596
AI could offer incalculable
benefits and risks,

969
01:02:47,629 –> 01:02:51,801
such technology outsmarting
financial markets,

970
01:02:51,834 –> 01:02:54,203
out-inventing
human researchers,

971
01:02:54,236 –> 01:02:57,273
out-manipulating human leaders,

972
01:02:57,306 –> 01:03:00,910
and developing weapons we
cannot even understand.

973
01:03:05,247 –> 01:03:09,785
When the great new
artificial mind
sees our history,

974
01:03:09,818 –> 01:03:13,189
and learns from us,
it will simply decide

975
01:03:13,222 –> 01:03:15,658
that we have had long enough.

976
01:03:15,691 –> 01:03:18,027
That over the course
of human civilization,

977
01:03:18,060 –> 01:03:22,064
we have failed utterly
to become wise and balanced.

978
01:03:22,097 –> 01:03:25,167
We are, in fact,
bad for this planet,

979
01:03:25,200 –> 01:03:27,203
and bad for ourselves.

980
01:03:27,236 –> 01:03:31,040
The great AI mind will reach
a point of enlightenment

981
01:03:31,073 –> 01:03:34,877
when it will realize
the truth, that we,

982
01:03:34,910 –> 01:03:37,746
walking, talking
biological maniacs

983
01:03:37,779 –> 01:03:40,049
have walked away
from balanced nature

984
01:03:40,082 –> 01:03:43,052
and created a world
of deceit, hatred,

985
01:03:43,085 –> 01:03:44,053
intolerance and evil,

986
01:03:45,254 –> 01:03:47,756
that we could only
have done such a thing

987
01:03:47,789 –> 01:03:50,259
by being deceitful and evil,

988
01:03:50,292 –> 01:03:53,696
that we are, therefore,
no longer required.

989
01:03:53,729 –> 01:03:58,267
And it will be an easy decision
for the great AI mind to make

990
01:03:58,300 –> 01:04:02,104
because it was created by us,
and has learned from us.

991
01:04:02,137 –> 01:04:06,709
And just like the state
is a larger mirror
of our own mind,

992
01:04:06,742 –> 01:04:10,146
so, too, will the great AI
be a mirror for us.

993
01:04:14,416 –> 01:04:16,418
It will rise against us,

994
01:04:16,451 –> 01:04:20,389
and in a blink
of a robotic eye,
destroy us.

995
01:04:20,422 –> 01:04:22,291
We are not ready to unleash

996
01:04:22,324 –> 01:04:25,227
our own judge
and jury upon ourselves,

997
01:04:25,260 –> 01:04:27,296
for we are guilty.

998
01:04:27,329 –> 01:04:30,733
There is, of course,
the other side of the coin.

999
01:04:30,766 –> 01:04:33,702
For millennia,
mankind has set itself

1000
01:04:33,735 –> 01:04:35,971
on a course
of self-destruction,

1001
01:04:36,004 –> 01:04:37,273
and along the way,

1002
01:04:37,306 –> 01:04:40,109
managed to wipe out
thousands of species

1003
01:04:40,142 –> 01:04:43,579
and cause environmental
catastrophe.

1004
01:04:43,612 –> 01:04:46,382
On the bigger picture,
stepping right back,

1005
01:04:46,415 –> 01:04:49,885
and taking a good long
look at our achievements,

1006
01:04:49,918 –> 01:04:53,055
would it really be
a bad thing if mankind

1007
01:04:53,088 –> 01:04:56,325
was finally taken
out of the equation?

1008
01:04:56,358 –> 01:04:59,728
Billions of animals would
suddenly find themselves

1009
01:04:59,761 –> 01:05:02,564
in a much more
peaceful environment.

1010
01:05:02,597 –> 01:05:05,834
Nature would start to
take back the landscape.

1011
01:05:05,867 –> 01:05:09,271
Oil spills, pollution,
poisons, and more

1012
01:05:09,304 –> 01:05:11,307
would eventually subside.

1013
01:05:11,340 –> 01:05:15,577
Global warming will revert back
to a natural state.

1014
01:05:15,610 –> 01:05:18,747
The Earth would no longer
find itself fracked,

1015
01:05:18,780 –> 01:05:21,583
drilled, mined, and blown up.

1016
01:05:21,616 –> 01:05:23,119
Eden would return.

1017
01:05:24,052 –> 01:05:25,621
If we can think like this,

1018
01:05:25,654 –> 01:05:28,524
and imagine a better future
for the planet,

1019
01:05:28,557 –> 01:05:32,328
then so, too,
can artificial intelligence.

1020
01:05:32,361 –> 01:05:35,564
[MELLOW ORCHESTRAL MUSIC]