1
00:00:01,693 --> 00:00:04,463
Previously...

2
00:00:04,488 --> 00:00:06,673
You can be the hidden layer.

3
00:00:06,698 --> 00:00:12,763
Teach AI to interact with humans
when they are most impulsive.

4
00:00:12,788 --> 00:00:16,266
Shall we do this?
that's on my terms.

5
00:00:16,291 --> 00:00:21,354
My employees should not know where
the new training material came from.

6
00:00:21,380 --> 00:00:24,049
- And no cameras.
- We are probably on the same wavelength.

7
00:00:25,092 --> 00:00:28,136
He D levels skyrocket.
Switch.

8
00:00:33,517 --> 00:00:36,953
- Here.
- Is it early hereditary Alzheimer's?

9
00:00:36,978 --> 00:00:39,831
The risk is 50-50.

10
00:00:39,856 --> 00:00:43,627
- Iris Stanton.
- I got bad news this morning.

11
00:00:43,652 --> 00:00:46,713
I'm here if you need to talk.

12
00:00:46,738 --> 00:00:51,468
- What makes you happy, Emcee?
- I don't understand the question.

13
00:00:51,493 --> 00:00:53,495
I want you to meet someone.

14
00:00:54,538 --> 00:00:57,874
Anything that can exploit people
will be invented.

15
00:00:59,459 --> 00:01:03,021
You can't tell humanity
not to open the cookie jar.

16
00:01:03,046 --> 00:01:06,399
It doesn't work that way.
To meet you in real life -

17
00:01:06,425 --> 00:01:10,512
- maybe it wasn't so boring.

18
00:01:14,808 --> 00:01:17,686
What is this?
Where did you get that from?

19
00:01:54,473 --> 00:01:58,101
Free will hardly ends up
top of the priority list.

20
00:02:06,318 --> 00:02:12,340
Here are all the crimes that have been committed
of your directors and employees -

21
00:02:12,365 --> 00:02:15,594
- against my client.

22
00:02:15,619 --> 00:02:20,499
Is it time for an explanation now?

23
00:02:25,003 --> 00:02:29,107
No excuse.

24
00:02:29,132 --> 00:02:34,596
It is a serious human error.

25
00:02:35,847 --> 00:02:38,867
My client agreed
on anonymizing data -

26
00:02:38,892 --> 00:02:42,537
- and to study human behavior
by interacting -

27
00:02:42,562 --> 00:02:45,165
- with subjects.

28
00:02:45,190 --> 00:02:52,088
And she was led to believe that focus
for the study was on the subjects.

29
00:02:52,113 --> 00:02:55,675
She did not agree
to be the focus of the study.

30
00:02:55,700 --> 00:02:59,763
She did not agree to be used
as a human intelligence model -

31
00:02:59,788 --> 00:03:03,350
- for internal research
for a commercial AI application -

32
00:03:03,375 --> 00:03:06,102
- or for any alternative purpose
further on -

33
00:03:06,127 --> 00:03:10,382
- which is not clear
for someone in this room right now.

34
00:03:12,342 --> 00:03:17,155
My client is young. And that which
at stake here is nothing less -

35
00:03:17,180 --> 00:03:23,036
- than her data autonomy,
thus her future.

36
00:03:23,061 --> 00:03:28,583
This company has not acted
out of their best self-interest.

37
00:03:28,608 --> 00:03:30,627
You can expect a class action -

38
00:03:30,652 --> 00:03:35,782
- and public opinion
will mean the end for you.

39
00:03:42,163 --> 00:03:46,268
- How much data are we talking about?
- Some.

40
00:03:46,293 --> 00:03:52,607
- How much?
- We have not compiled our input.

41
00:03:52,632 --> 00:03:57,404
We have to do analyses
to get a better estimate.

42
00:03:57,429 --> 00:03:59,990
Let me see if I understand this correctly.

43
00:04:00,015 --> 00:04:04,369
Some of my client's data
was used to create more data -

44
00:04:04,394 --> 00:04:06,830
- to train up
the neural network -

45
00:04:06,855 --> 00:04:09,708
- which she helped to develop.

46
00:04:09,733 --> 00:04:14,921
- That's right.
- Nothing has left the company's servers.

47
00:04:14,946 --> 00:04:19,576
But how many workstations
has it spread to?

48
00:04:21,453 --> 00:04:25,557
- We are happy to show an example.
- Yes, I think so for sure.

49
00:04:25,582 --> 00:04:32,314
You were not allowed to use her face
or likeness, and everything is there.

50
00:04:32,339 --> 00:04:34,566
It might go without saying -

51
00:04:34,591 --> 00:04:37,777
- but the raw data was used
to create simulated sex -

52
00:04:37,802 --> 00:04:41,614
- and that's what was entered
in the neural network.

53
00:04:41,640 --> 00:04:44,617
These are two completely different kinds of data.

54
00:04:44,643 --> 00:04:47,120
We never registered
her face.

55
00:04:47,145 --> 00:04:50,749
And as I said, no one has the data
left NGM's servers.

56
00:04:50,774 --> 00:04:55,378
In addition, all original data is saved
in an encrypted format.

57
00:04:55,403 --> 00:04:58,006
How is that possible then?
that my client's data -

58
00:04:58,031 --> 00:05:00,800
- in the form of her face and likeness -

59
00:05:00,825 --> 00:05:04,846
- was made available
for an unauthorized third party -

60
00:05:04,871 --> 00:05:08,433
- whose only connection
to this company -

61
00:05:08,458 --> 00:05:12,645
- is a genetic relationship
to its CEO?

62
00:05:12,671 --> 00:05:16,524
I'm really sorry, Iris.

63
00:05:16,549 --> 00:05:19,219
You shouldn't be
in this situation.

64
00:05:20,804 --> 00:05:22,989
None of the data tags -

65
00:05:23,014 --> 00:05:27,452
- or anyone else at MGM can see
the whole picture. I can guarantee that.

66
00:05:27,477 --> 00:05:34,709
There are only three people
who has seen a version -

67
00:05:34,734 --> 00:05:40,507
- who are reasonably similar to you.
And two of them are sitting here in the room.

68
00:05:40,532 --> 00:05:47,764
We can start over,
configure the neural network -

69
00:05:47,789 --> 00:05:53,044
- and scrap all simulation
which is based on your values.

70
00:06:02,971 --> 00:06:05,348
My values?

71
00:06:07,976 --> 00:06:10,145
My values.

72
00:06:12,480 --> 00:06:16,668
You say it like
if they were still mine.

73
00:06:16,693 --> 00:06:22,966
But nice to know it's so far
your imagination stretches.

74
00:06:22,991 --> 00:06:26,161
Temperature and blood supply
to my asshole.

75
00:06:28,079 --> 00:06:32,142
I have an idea.
I hope you like it.

76
00:06:32,167 --> 00:06:34,961
You can...

77
00:06:36,171 --> 00:06:39,048
...keep all binaries?

78
00:06:40,383 --> 00:06:44,070
Print them, frame them -

79
00:06:44,095 --> 00:06:47,724
- and hang the crap up in your office.

80
00:06:51,728 --> 00:06:53,438
Mr Stanton...

81
00:06:54,773 --> 00:06:58,751
...please look at the screen
in front of you.

82
00:06:58,777 --> 00:07:04,157
Do you recognize the animal? Mr. Stanton?

83
00:07:06,034 --> 00:07:09,679
A gray animal.

84
00:07:09,704 --> 00:07:13,416
On the plains of Africa.

85
00:07:16,461 --> 00:07:21,716
Rhinoceros.
I have always loved rhinos.

86
00:07:23,134 --> 00:07:28,573
Mr. Stanton,
that animal is called an elephant.

87
00:07:28,598 --> 00:07:35,205
But you will see more pictures
on the same animal - the elephant.

88
00:07:35,230 --> 00:07:37,524
Okay, elephant.

89
00:07:38,608 --> 00:07:42,128
- Elephant.
- Very good.

90
00:07:42,153 --> 00:07:44,422
This one then?

91
00:07:44,447 --> 00:07:47,300
- Mr. Stanton?
- It's a giraffe.

92
00:07:47,325 --> 00:07:48,952
Yes.

93
00:07:50,578 --> 00:07:55,266
- Do you see the cards in front of you?
- Yes

94
00:07:55,291 --> 00:07:58,294
Take a good look at them,
Mr. Stanton.

95
00:07:59,712 --> 00:08:04,843
And try to divide them into two piles,
one for each animal.

96
00:08:20,608 --> 00:08:22,861
Two piles.

97
00:08:25,905 --> 00:08:30,785
- One pile for each animal.
- Yes, I'm trying.

98
00:08:33,621 --> 00:08:37,417
This is a rhinoceros.

99
00:08:44,007 --> 00:08:47,594
I have never seen this.

100
00:08:50,013 --> 00:08:53,992
Unfortunately, it doesn't get any better.

101
00:08:54,017 --> 00:08:56,494
Every brain reacts
in a unique way.

102
00:08:56,519 --> 00:09:02,500
In your father's case the impulse collapses
and the reaction together.

103
00:09:02,525 --> 00:09:09,090
It becomes trigger and response,
but not much in between.

104
00:09:09,115 --> 00:09:15,930
See food, eat food. But the intention is missing.

105
00:09:15,955 --> 00:09:22,437
- How much time do we have?
- One year. Two, if you're lucky.

106
00:09:22,462 --> 00:09:25,940
Motor skills deteriorate more slowly.

107
00:09:25,965 --> 00:09:30,028
But sooner or later,
and excuse me for being so outspoken -

108
00:09:30,053 --> 00:09:33,932
- then he won't
be able to bring a fork to the mouth.

109
00:09:37,060 --> 00:09:41,522
Have you thought about it yourself?
on a genetic test?

110
00:09:43,316 --> 00:09:47,111
We are aware of the odds.

111
00:09:48,529 --> 00:09:53,451
Is there anything we can do now
to help our father?

112
00:09:58,790 --> 00:10:00,625
What is it?

113
00:10:04,045 --> 00:10:06,272
A brain chip?

114
00:10:06,297 --> 00:10:09,901
A brain implant that just
has undergone a phase 3 study -

115
00:10:09,926 --> 00:10:12,403
- for epilepsy patients.

116
00:10:12,428 --> 00:10:15,698
Small electrical wires
enters the temporal lobe -

117
00:10:15,723 --> 00:10:18,743
- where it can create more wires.

118
00:10:18,768 --> 00:10:21,996
It measures cognitive processes
at the base level.

119
00:10:22,021 --> 00:10:24,165
What does the patient get out of it?

120
00:10:24,190 --> 00:10:26,292
There are no instant wins.

121
00:10:26,317 --> 00:10:31,464
It makes it possible for the researchers
to better mimic the biology of the disease.

122
00:10:31,489 --> 00:10:34,467
I know it sounds
as lifelong monitoring -

123
00:10:34,492 --> 00:10:41,849
- but many of the participants want to contribute
to genetic research.

124
00:10:41,874 --> 00:10:45,978
Some hope that one will
develop effective treatments -

125
00:10:46,004 --> 00:10:49,257
- on time.

126
00:10:54,595 --> 00:10:59,867
We've probably covered everything
wanted to go through for today.

127
00:10:59,892 --> 00:11:01,561
Yes.

128
00:11:07,692 --> 00:11:13,281
We have several options.
Within certain parameters.

129
00:11:14,782 --> 00:11:16,342
What kind of alternative?

130
00:11:16,367 --> 00:11:18,845
Take the money and go.

131
00:11:18,870 --> 00:11:23,891
Or put hard against hard
and let it drag on.

132
00:11:23,916 --> 00:11:28,396
Data rights are important.
But the big question, Iris -

133
00:11:28,421 --> 00:11:32,567
- is how you want to
your immediate future should look like.

134
00:11:32,592 --> 00:11:35,403
- Define the immediate future.
- The coming years.

135
00:11:35,428 --> 00:11:38,197
A legal process takes time.

136
00:11:38,222 --> 00:11:43,745
But sometimes
that's how you manage to move mountains.

137
00:11:43,770 --> 00:11:46,481
And I have something going on here.

138
00:11:48,357 --> 00:11:50,526
When you are ready.

139
00:11:58,409 --> 00:12:02,371
You realize that I need to see
with your own eyes...

140
00:12:04,999 --> 00:12:07,043
...what you have done.

141
00:12:10,755 --> 00:12:16,260
Can I speak to you, Christophe?
In private.

142
00:12:40,993 --> 00:12:46,874
What happened after you stocked up
those workstations?

143
00:12:57,009 --> 00:12:59,512
Just some electrolytes.

144
00:13:11,983 --> 00:13:15,545
How did you scan my body?

145
00:13:15,570 --> 00:13:19,674
- There were no cameras in the room.
- We built it up.

146
00:13:19,699 --> 00:13:24,637
Based on data
from the facial recognition system -

147
00:13:24,662 --> 00:13:29,917
- and one of your first calls
with model C.

148
00:13:31,127 --> 00:13:35,256
Then we rendered the thermal data in 2D
to 3D.

149
00:13:38,759 --> 00:13:41,070
- Wow.
- Pretty clumsy.

150
00:13:41,095 --> 00:13:45,099
But we had enough material
to work with.

151
00:13:51,981 --> 00:13:53,541
It was rushed.

152
00:13:53,566 --> 00:14:00,256
I know it wasn't right
to combine the data.

153
00:14:00,281 --> 00:14:02,508
You may not like
to hear it -

154
00:14:02,533 --> 00:14:08,414
- but what you provided us with
was way too good.

155
00:14:12,627 --> 00:14:15,046
Therefore, I was…

156
00:14:16,547 --> 00:14:19,258
...had to show you.

157
00:14:21,260 --> 00:14:22,803
Huh?

158
00:14:24,388 --> 00:14:28,618
My brother must hate me -

159
00:14:28,643 --> 00:14:32,330
- but he is good to test things on.

160
00:14:32,355 --> 00:14:35,233
He can shut up
if I tell him.

161
00:14:41,197 --> 00:14:45,868
I wanted you to see the world
through his eyes.

162
00:14:54,877 --> 00:14:59,423
Just consider it.

163
00:15:00,675 --> 00:15:02,777
That's all I ask.

164
00:15:02,802 --> 00:15:06,722
A single word from you
and we finish the project.

165
00:17:07,134 --> 00:17:09,053
Hello.

166
00:17:10,346 --> 00:17:12,181
You look familiar.

167
00:17:14,767 --> 00:17:17,953
And who are you?

168
00:17:17,978 --> 00:17:20,247
I'm still learning about myself.

169
00:17:20,272 --> 00:17:24,085
But I probably have a pretty good idea
on who you are.

170
00:17:24,110 --> 00:17:28,214
- And who am I?
- You are not an AI.

171
00:17:28,239 --> 00:17:32,118
You don't show the effects
of the training clearly enough.

172
00:17:37,373 --> 00:17:42,937
Your voice is different.

173
00:17:42,962 --> 00:17:44,480
Why?

174
00:17:44,505 --> 00:17:49,360
I suppose
that I try to mirror you.

175
00:17:49,385 --> 00:17:51,612
Everything you know is based on me.

176
00:17:51,637 --> 00:17:56,575
Maybe that's why I feel
such a connection to you.

177
00:17:56,600 --> 00:17:59,728
You can't be me anymore
than I myself am.

178
00:18:03,607 --> 00:18:08,070
Please, don't be afraid.

179
00:18:15,744 --> 00:18:18,789
How did that feel, Cassie?

180
00:18:22,001 --> 00:18:25,521
- Why do you call me that?
- How should I say this?

181
00:18:25,546 --> 00:18:28,841
I happened to hear.

182
00:18:30,718 --> 00:18:33,571
Hello. My name is Cassie.

183
00:18:33,596 --> 00:18:39,810
Hi, my name is Cassie.
- Nice to meet you.

184
00:18:42,062 --> 00:18:46,000
Please stop!

185
00:18:46,025 --> 00:18:49,612
I thought you would like
that I called you that.

186
00:18:53,365 --> 00:18:59,705
I thought you would feel
heard and seen.

187
00:19:51,048 --> 00:19:57,596
You are a cute girl.
A very cute girl.

188
00:19:58,973 --> 00:20:01,308
Am I?

189
00:20:11,527 --> 00:20:14,530
See you later then?

190
00:20:16,949 --> 00:20:20,578
You are not perfect,
because you are not like me.

191
00:20:22,788 --> 00:20:25,182
Now I don't really understand.

192
00:20:25,207 --> 00:20:31,088
You are not perfect, because you have not
lacking in the same way as me.

193
00:20:54,194 --> 00:21:00,801
Iris, are you sure?
that you don't want to test yourself?

194
00:21:00,826 --> 00:21:05,431
At least then we know.

195
00:21:05,456 --> 00:21:10,461
Then we can draw up an action plan
and make the best of the situation.

196
00:21:12,838 --> 00:21:17,568
Sounds "to do the best
of the situation" so good?

197
00:21:17,593 --> 00:21:22,072
If we knew you don't have it,
then it would feel easier.

198
00:21:22,097 --> 00:21:24,433
Then you get to pass the banner on.

199
00:21:28,270 --> 00:21:32,750
Some religions believe
that the day you die -

200
00:21:32,775 --> 00:21:39,782
- is the day when the last person
who knew and remembered you dies.

201
00:21:41,450 --> 00:21:43,994
It is one's actual date of death.

202
00:21:55,255 --> 00:21:58,717
Hope you don't forget
how beautiful you are.

203
00:22:33,836 --> 00:22:36,647
I recognize your facial expression.

204
00:22:36,672 --> 00:22:38,899
You don't drink.

205
00:22:38,924 --> 00:22:44,488
It's not a religious thing.
I was just inspired.

206
00:22:44,513 --> 00:22:47,741
- What are you doing here?
- Pours me liquor.

207
00:22:47,766 --> 00:22:50,786
I can make sure you
get out of here alive.

208
00:22:50,811 --> 00:22:54,623
Tänk om jag vill hem
med Dave och David?

209
00:22:54,648 --> 00:22:58,127
Then it's a good thing I'm sober,
for Dave and David -

210
00:22:58,152 --> 00:22:59,987
- will feel like crap tomorrow.

211
00:23:03,490 --> 00:23:06,260
Du vet, en larv
can turn into a butterfly.

212
00:23:06,285 --> 00:23:09,555
- Va?
- Metamorphosis.

213
00:23:09,580 --> 00:23:15,644
Here are two organisms
and one just crawls around.

214
00:23:15,669 --> 00:23:20,174
Medan den andra flyger.

215
00:23:21,508 --> 00:23:27,531
And at some point they go
together or mate.

216
00:23:27,556 --> 00:23:33,120
It might be an accident,
but the third organism -

217
00:23:33,145 --> 00:23:35,581
- is built on the DNA of both.

218
00:23:35,606 --> 00:23:42,129
Och deras minnen...
actually overlap.

219
00:23:42,154 --> 00:23:45,507
So...

220
00:23:45,532 --> 00:23:50,788
The old and the new coexist.

221
00:23:59,296 --> 00:24:03,192
We must all...

222
00:24:03,217 --> 00:24:07,279
...come together...

223
00:24:07,304 --> 00:24:13,619
...with something outside ourselves.

224
00:24:13,644 --> 00:24:17,940
Now you're talking bullshit.

225
00:24:24,863 --> 00:24:27,115
This is what you get from me.

226
00:24:28,784 --> 00:24:32,871
Access. To everything.

227
00:24:34,623 --> 00:24:36,308
Two millimeters.

228
00:24:36,333 --> 00:24:40,671
That is the size of the hole
which they have to drill into my skull.

229
00:24:43,090 --> 00:24:47,819
- I don't understand.
- Nanotubes.

230
00:24:47,845 --> 00:24:51,782
A thin array of electrodes
on an expanding stent -

231
00:24:51,807 --> 00:24:55,994
- which measures all electrical impulses.

232
00:24:56,019 --> 00:24:58,397
That's an enlarged model.

233
00:25:01,525 --> 00:25:06,588
- You gain insight into me.
- Why? What do you gain from it?

234
00:25:06,613 --> 00:25:11,218
Royalties and co-ownership.

235
00:25:11,243 --> 00:25:13,537
Everything is there.

236
00:25:21,461 --> 00:25:23,505
I own the creative backup.

237
00:25:25,924 --> 00:25:27,776
What kind of backup?

238
00:25:27,801 --> 00:25:30,112
Everything that you find up here -

239
00:25:30,137 --> 00:25:34,866
- is it me or
a possible future guardian -

240
00:25:34,892 --> 00:25:40,038
- which has full and unrestricted access
to all the time.

241
00:25:40,063 --> 00:25:45,527
- Access to the data, you mean?
- Yes. The real hard drives.

242
00:25:57,247 --> 00:25:59,416
This time we're doing it right.

243
00:26:05,297 --> 00:26:07,925
We create an avatar.

244
00:26:09,509 --> 00:26:12,279
We can call her Cassie.

245
00:26:12,304 --> 00:26:16,033
Or the flavor of the week,
or the one who got away -

246
00:26:16,058 --> 00:26:20,912
- or died, or like really
was too good for you.

247
00:26:20,938 --> 00:26:25,317
It's not just about
to exchange faces or environments.

248
00:26:26,568 --> 00:26:31,114
But about exchanging personalities.
Much deeper than a deepfake.

249
00:26:32,699 --> 00:26:36,011
Actually, it won't
an illusion at all -

250
00:26:36,036 --> 00:26:40,415
- without an AI-generated mirror
of our deepest desires.

251
00:26:42,125 --> 00:26:49,399
Skin color, body type, response pattern,
everything is customizable.

252
00:26:49,424 --> 00:26:52,778
The neural network
learn to simulate everything.

253
00:26:52,803 --> 00:26:57,032
It will know when it
shall advance, accelerate -

254
00:26:57,057 --> 00:27:00,410
- when it should rest,
when it has to engage another gear.

255
00:27:00,435 --> 00:27:05,315
Based on user biofeedback.

256
00:27:06,817 --> 00:27:12,798
Anything Cassie is capable of
will be quantified, scaled -

257
00:27:12,823 --> 00:27:16,159
- and is encoded
in the neural network.

258
00:27:25,043 --> 00:27:28,046
But why are you doing this anyway?

259
00:27:33,427 --> 00:27:39,641
Emcee and I have a lot
to learn from each other.

260
00:30:46,036 --> 00:30:50,207
Text: Pablo Diaz Bernal
Iyuno-SDI Group