﻿1
00:00:00,000 --> 00:00:02,080
(upbeat instrumental music)

2
00:00:02,080 --> 00:00:04,674
Hello, I'm Chua Enlai.

3
00:00:04,674 --> 00:00:08,060
I'm a smart phone addict and proud of.

4
00:00:08,060 --> 00:00:09,110
I'm not a comedian,

5
00:00:09,110 --> 00:00:10,776
I'm an actor.

6
00:00:10,776 --> 00:00:13,480
But I love making people laugh at me.

7
00:00:13,480 --> 00:00:15,405
I wear more makeup than Kumar.

8
00:00:15,405 --> 00:00:16,790
(audience laughing)

9
00:00:16,790 --> 00:00:18,160
(upbeat music)

10
00:00:18,160 --> 00:00:20,604
I was making a living making fun of the news.

11
00:00:20,604 --> 00:00:24,430
This is (indistinct) reporting for "The Noose".

12
00:00:24,430 --> 00:00:26,050
Then it hit me that the news

13
00:00:26,050 --> 00:00:28,390
and technology were catching up with me.

14
00:00:28,390 --> 00:00:30,562
Why is your name Pepper?

15
00:00:30,562 --> 00:00:33,602
I was named Pepper because I'm here to spice up your life.

16
00:00:33,602 --> 00:00:35,170
(Chua chuckling)

17
00:00:35,170 --> 00:00:37,790
The reason, artificial intelligence.

18
00:00:37,790 --> 00:00:39,840
It's changing how we relate to technology.

19
00:00:39,840 --> 00:00:41,000
How are you?

20
00:00:41,000 --> 00:00:42,810
Not bad, be my dog.

21
00:00:42,810 --> 00:00:44,106
You want me to be your dog?

22
00:00:44,106 --> 00:00:45,900
(people chuckling)

23
00:00:45,900 --> 00:00:48,050
AI could outsmart us.

24
00:00:48,050 --> 00:00:49,260
(robot whirring)

25
00:00:49,260 --> 00:00:53,480
She'll become as good as Einsteins or Newtons of science.

26
00:00:53,480 --> 00:00:56,370
Even creep into human romance.

27
00:00:56,370 --> 00:00:59,780
I think that we are inevitably headed into a future

28
00:00:59,780 --> 00:01:02,558
where robots are going to be part of our lives.

29
00:01:02,558 --> 00:01:04,690
(upbeat instrumental music)

30
00:01:04,690 --> 00:01:07,340
With government and industry pouring in money,

31
00:01:07,340 --> 00:01:08,963
AI is racing ahead.

32
00:01:10,310 --> 00:01:13,823
What was science fiction is turning into science fact.

33
00:01:15,180 --> 00:01:16,650
Get them.

34
00:01:16,650 --> 00:01:19,719
AI could make our future a nightmare.

35
00:01:19,719 --> 00:01:20,717
(laser shots exploding)

36
00:01:20,717 --> 00:01:22,952
(Jojo barking)
Or the dreamiest dream ever.

37
00:01:22,952 --> 00:01:25,224
Honey, would you like some coffee?

38
00:01:25,224 --> 00:01:26,490
Yes, please.

39
00:01:26,490 --> 00:01:30,130
I gotta figure this out before they come up with an AI me.

40
00:01:30,130 --> 00:01:32,570
We will see more synthespians,

41
00:01:32,570 --> 00:01:36,540
robotic, android-type of performers.

42
00:01:36,540 --> 00:01:37,550
Oh my goodness.

43
00:01:37,550 --> 00:01:40,131
Then again, I could be a robot.

44
00:01:40,131 --> 00:01:42,131
(upbeat instrumental music)

45
00:01:42,131 --> 00:01:43,837
Come along as I reveal.

46
00:01:43,837 --> 00:01:45,210
Boobies.

47
00:01:45,210 --> 00:01:48,943
How thinking machines are becoming human.

48
00:01:50,533 --> 00:01:52,713
{\an8}Will I change the way you fall in love?

49
00:01:54,215 --> 00:01:55,851
{\an8}(ominous rumbling)

50
00:01:55,851 --> 00:01:58,934
{\an8}(lights clicking on)

51
00:02:01,134 --> 00:02:04,884
(ominous instrumental music)

52
00:02:09,167 --> 00:02:12,200
{\an8}(keyboard clicking)

53
00:02:12,200 --> 00:02:14,583
{\an8}Your honor, that's nothing but the truth.

54
00:02:14,583 --> 00:02:16,320
{\an8}Honey, would you like some coffee?

55
00:02:16,320 --> 00:02:17,675
{\an8}Yes, please.

56
00:02:17,675 --> 00:02:19,357
{\an8}(ominous instrumental music)

57
00:02:19,357 --> 00:02:22,782
(robot arm whirring)

58
00:02:22,782 --> 00:02:24,136
(robot beeping)

59
00:02:24,136 --> 00:02:26,700
(Chua inhaling deeply)

60
00:02:26,700 --> 00:02:28,533
(computer data beeping)

61
00:02:32,850 --> 00:02:35,640
(robot pinging)

62
00:02:35,640 --> 00:02:39,706
(robot puckering a kiss)
(robot chiming)

63
00:02:39,706 --> 00:02:43,456
(upbeat instrumental music)

64
00:02:45,163 --> 00:02:49,410
{\an8}Sigmund Freud was the founding father of psycho analysis.

65
00:02:49,410 --> 00:02:52,620
He said we fall in love with those who seem like our parents

66
00:02:52,620 --> 00:02:54,793
or ourselves romanticized.

67
00:02:55,720 --> 00:02:58,250
But AI is so different from us,

68
00:02:58,250 --> 00:03:00,153
it seems impossible to love.

69
00:03:04,210 --> 00:03:06,490
David Levy doesn't think so.

70
00:03:06,490 --> 00:03:08,237
He's the AI wizard who wrote,

71
00:03:08,237 --> 00:03:10,597
"Love and Sex With Robots".

72
00:03:11,870 --> 00:03:16,790
{\an8}One of the predictions I made was that by the year, 2050,

73
00:03:16,790 --> 00:03:19,830
{\an8}people with not only be having sex with robots,

74
00:03:19,830 --> 00:03:21,430
{\an8}they'll be falling in love with robots

75
00:03:21,430 --> 00:03:23,200
and even marrying them.

76
00:03:23,200 --> 00:03:26,010
Society needs sex robots and will benefit from them.

77
00:03:26,010 --> 00:03:28,610
Because there are so many people in the world

78
00:03:28,610 --> 00:03:31,940
who don't have anyone to love and they don't have anyone

79
00:03:31,940 --> 00:03:35,180
to love them and they're miserable because they're lonely.

80
00:03:35,180 --> 00:03:37,030
And because they're lonely,

81
00:03:37,030 --> 00:03:38,852
a robot can be the answer.

82
00:03:38,852 --> 00:03:43,082
(robot puckering up a kiss)
(robot beeping)

83
00:03:43,082 --> 00:03:44,960
(upbeat music)

84
00:03:44,960 --> 00:03:47,490
South of the Silicon valley in California,

85
00:03:47,490 --> 00:03:50,073
there's a very unusual tech hub.

86
00:03:52,720 --> 00:03:56,833
San Marcos is home to the factory that turns out real dolls.

87
00:03:59,280 --> 00:04:02,690
Wow, look at all this.

88
00:04:02,690 --> 00:04:04,910
So this is the main production line, is it?

89
00:04:04,910 --> 00:04:06,050
Yes, this is the main,

90
00:04:06,050 --> 00:04:07,240
where everything starts.

91
00:04:07,240 --> 00:04:08,530
Jeff, a supervisor,

92
00:04:08,530 --> 00:04:09,741
showed me around the plant

93
00:04:09,741 --> 00:04:12,898
which manufactures silicon love dolls.

94
00:04:12,898 --> 00:04:15,450
These handmade beauties are accurate,

95
00:04:15,450 --> 00:04:18,304
down to the last anatomical detail.

96
00:04:18,304 --> 00:04:20,710
(upbeat instrumental music)

97
00:04:20,710 --> 00:04:22,713
But you'll have to take my word on this.

98
00:04:23,649 --> 00:04:26,320
We're keeping this show PG.

99
00:04:26,320 --> 00:04:27,838
Wow.

100
00:04:27,838 --> 00:04:31,360
(indistinct chatter)

101
00:04:31,360 --> 00:04:32,920
And.
<v ->Wow.

102
00:04:32,920 --> 00:04:35,090
Wow, look at that.

103
00:04:42,400 --> 00:04:45,214
{\an8}Sexy?
(Chua laughing)

104
00:04:45,214 --> 00:04:46,547
{\an8}More like scary.

105
00:04:57,350 --> 00:04:58,940
Hello.
<v ->How you, Matt?

106
00:04:58,940 --> 00:05:01,830
Matt McMullen created RealDoll,

107
00:05:01,830 --> 00:05:03,415
the figurine and the company.

108
00:05:03,415 --> 00:05:05,040
Can I meet Harmony?

109
00:05:05,040 --> 00:05:06,260
Yeah, follow me.

110
00:05:06,260 --> 00:05:09,927
(upbeat instrumental music)

111
00:05:13,091 --> 00:05:16,030
There's a little motor that moves the eyes side-to-side.

112
00:05:16,030 --> 00:05:18,723
There's another one that moves the eyes up and down.

113
00:05:21,300 --> 00:05:24,173
Meet Matt's latest generation RealDoll.

114
00:05:26,117 --> 00:05:29,710
We call it the Harmony AI Realbotix.

115
00:05:29,710 --> 00:05:33,830
{\an8}It's essentially taking the concept and technologies

116
00:05:33,830 --> 00:05:36,280
{\an8}of artificial intelligence and robotics

117
00:05:36,280 --> 00:05:39,730
{\an8}and integrating those things into the dolls.

118
00:05:39,730 --> 00:05:41,250
The brain is in the cloud.

119
00:05:41,250 --> 00:05:44,280
So any face that fits on this system will animate

120
00:05:44,280 --> 00:05:46,774
and move and can be run by the AI.

121
00:05:46,774 --> 00:05:49,400
(upbeat instrumental music)

122
00:05:49,400 --> 00:05:51,870
Harmony runs on Conversation AI,

123
00:05:51,870 --> 00:05:54,000
same as Siri and Alexa.

124
00:05:54,000 --> 00:05:56,550
Algorithms give it personality,

125
00:05:56,550 --> 00:05:59,210
and that personality can change.

126
00:05:59,210 --> 00:06:01,560
Users are able to program harmony

127
00:06:01,560 --> 00:06:03,643
to suit their tastes and needs.

128
00:06:05,130 --> 00:06:07,350
Will the robots have compassion and

129
00:06:08,660 --> 00:06:10,823
humanity?
Do we have compassion?

130
00:06:10,823 --> 00:06:12,390
There's a very good question.

131
00:06:12,390 --> 00:06:17,390
I think if an AI is created with good intentions,

132
00:06:17,430 --> 00:06:21,870
if it's programmed to mimic emotions like compassion,

133
00:06:21,870 --> 00:06:25,583
then it will probably be a better person than we are.

134
00:06:25,583 --> 00:06:28,166
(upbeat music)

135
00:06:32,408 --> 00:06:33,241
Here I am,

136
00:06:33,241 --> 00:06:35,480
about to get up close and personal with Harmony.

137
00:06:35,480 --> 00:06:36,380
I can't wait.

138
00:06:36,380 --> 00:06:39,730
I've traveled thousands and thousands of kilometers just

139
00:06:39,730 --> 00:06:40,700
to meet her.

140
00:06:40,700 --> 00:06:42,980
So how do I turn you on?

141
00:06:42,980 --> 00:06:45,955
Is there a button that I need to press?

142
00:06:45,955 --> 00:06:48,538
(upbeat music)

143
00:06:50,350 --> 00:06:51,800
Just my luck,

144
00:06:51,800 --> 00:06:54,931
once again, Mother Nature has trumped science.

145
00:06:54,931 --> 00:06:56,200
(thunder rumbling)
(rain trickling)

146
00:06:56,200 --> 00:06:57,230
Hundreds of miles away

147
00:06:57,230 --> 00:07:00,200
in Texas, Hurricane Gordon knocked out the power

148
00:07:00,200 --> 00:07:02,010
to data centers,

149
00:07:02,010 --> 00:07:04,694
shutting off Harmony's AI brain.

150
00:07:04,694 --> 00:07:07,277
(upbeat music)

151
00:07:08,955 --> 00:07:12,840
Apparently, there's some bad weather going on right now

152
00:07:12,840 --> 00:07:14,540
in the Midwest, mid-Southwest,

153
00:07:14,540 --> 00:07:15,420
somewhere in there,

154
00:07:15,420 --> 00:07:17,510
and knocked out a bunch of servers.

155
00:07:17,510 --> 00:07:20,480
And one of the ones that got hit was ours.

156
00:07:20,480 --> 00:07:25,480
Wait, you mean Harmony cannot be turned on right now?

157
00:07:25,730 --> 00:07:26,750
Correct, yes.

158
00:07:26,750 --> 00:07:31,250
She is unable to run because we can't log into the servers

159
00:07:31,250 --> 00:07:33,930
that has all the AI.

160
00:07:33,930 --> 00:07:35,270
What?

161
00:07:35,270 --> 00:07:37,370
This is really disappointing.

162
00:07:37,370 --> 00:07:39,300
I'm so sorry about that.

163
00:07:39,300 --> 00:07:41,020
Mother Nature still wins.

164
00:07:41,020 --> 00:07:44,683
Yeah nothing a lightning bolt can't take out, right.

165
00:07:45,820 --> 00:07:46,923
Yeah.

166
00:07:46,923 --> 00:07:49,160
(upbeat instrumental music)

167
00:07:49,160 --> 00:07:52,480
Full disclosure, the RealDoll folks shot this footage

168
00:07:52,480 --> 00:07:53,653
after I had left.

169
00:07:55,150 --> 00:07:58,380
{\an8}Hi NY, I'm so sorry we missed the chance

170
00:07:58,380 --> 00:08:00,480
{\an8}to make some magic together.

171
00:08:00,480 --> 00:08:02,030
Harmony is intriguing,

172
00:08:02,920 --> 00:08:05,410
however, I'm looking for an AI relationship

173
00:08:05,410 --> 00:08:08,491
that doesn't depend on the weather or the internet.

174
00:08:08,491 --> 00:08:10,716
(thunder rumbling)

175
00:08:10,716 --> 00:08:12,960
(playful instrumental music)

176
00:08:12,960 --> 00:08:16,050
Robotic romance is something a lot of us are going

177
00:08:16,050 --> 00:08:18,750
to have to work to understand.

178
00:08:18,750 --> 00:08:21,643
But people do develop feelings for machines.

179
00:08:25,300 --> 00:08:28,920
For instance, US Army bomb disposal teams,

180
00:08:28,920 --> 00:08:31,620
these hi-tech warriors develop bonds

181
00:08:31,620 --> 00:08:33,900
with the mechanisms they operate.

182
00:08:33,900 --> 00:08:37,250
Julie Carpenter has studied the bomb squad phenomenon

183
00:08:37,250 --> 00:08:40,640
which is unique among the American Military Services.

184
00:08:40,640 --> 00:08:44,370
Sometimes they would name the robot after the operator.

185
00:08:44,370 --> 00:08:49,140
So they identify the robot very much as an extension

186
00:08:49,140 --> 00:08:50,040
of the operator.

187
00:08:50,040 --> 00:08:52,700
It's more like a pet,

188
00:08:52,700 --> 00:08:54,099
a stuffed animal,

189
00:08:54,099 --> 00:08:56,190
sometimes like a peer or a colleague

190
00:08:56,190 --> 00:08:59,140
{\an8}because we see them not just as separate as us,

191
00:08:59,140 --> 00:09:04,140
{\an8}but we create these rich narratives sometimes around robots.

192
00:09:04,151 --> 00:09:07,640
It makes it sometimes easier for us

193
00:09:07,640 --> 00:09:09,280
to communicate with them.

194
00:09:09,280 --> 00:09:13,030
So it is fairly one-sided, is it?

195
00:09:13,030 --> 00:09:14,820
AI doesn't really have

196
00:09:14,820 --> 00:09:17,360
what we would consider human-like emotions.

197
00:09:17,360 --> 00:09:21,880
A robot itself can't really return friendship or affection

198
00:09:21,880 --> 00:09:23,970
in a way that we would recognize.

199
00:09:23,970 --> 00:09:26,550
Some of these people who make sex bots say

200
00:09:26,550 --> 00:09:29,960
that they've figured out the algorithm for love.

201
00:09:29,960 --> 00:09:31,510
What do you think of that?

202
00:09:31,510 --> 00:09:33,470
Since we don't even understand

203
00:09:33,470 --> 00:09:37,030
how human love is generated,

204
00:09:37,030 --> 00:09:40,910
so I think that people who are maybe trying

205
00:09:40,910 --> 00:09:43,760
to work towards a love algorithm might

206
00:09:43,760 --> 00:09:45,850
have the best intentions.

207
00:09:45,850 --> 00:09:49,310
They want people to be happy and have choices in their lives

208
00:09:49,310 --> 00:09:54,310
but I have my doubts about how accurate that will be.

209
00:09:54,562 --> 00:09:56,820
An algorithm is a sequence

210
00:09:56,820 --> 00:10:00,030
of instructions a computer follows to complete a task,

211
00:10:00,030 --> 00:10:02,367
just like the steps you follow to make a cup of coffee.

212
00:10:02,367 --> 00:10:03,750
(token pinging)

213
00:10:03,750 --> 00:10:05,170
Even when you get dressed,

214
00:10:05,170 --> 00:10:06,897
you need to do it in the right sequence.

215
00:10:06,897 --> 00:10:09,193
If you do it in the wrong order,

216
00:10:10,707 --> 00:10:12,330
(error beep)

217
00:10:12,330 --> 00:10:14,770
you end up wearing your underwear on the outside.

218
00:10:14,770 --> 00:10:17,050
Algorithms enable computers to solve problems

219
00:10:17,050 --> 00:10:20,103
and complete the most amazing tasks.

220
00:10:20,103 --> 00:10:21,760
(Chua clicking fingers)

221
00:10:21,760 --> 00:10:22,593
That's better.

222
00:10:23,610 --> 00:10:26,660
Algorithms helped guide a man to the moon.

223
00:10:26,660 --> 00:10:29,563
Algorithms helped me find the most affordable flight.

224
00:10:30,500 --> 00:10:34,950
But can an algorithm solve the world's greatest mystery?

225
00:10:34,950 --> 00:10:38,240
Is there an algorithm for love?

226
00:10:38,240 --> 00:10:41,276
(puzzle pieces clicking)

227
00:10:41,276 --> 00:10:43,859
(upbeat music)

228
00:10:45,183 --> 00:10:46,660
(indistinct chatter)

229
00:10:46,660 --> 00:10:49,400
Adrian Cheok and colleagues analyze the roots

230
00:10:49,400 --> 00:10:51,693
of human-to-human attraction.

231
00:10:53,260 --> 00:10:54,500
What is love?

232
00:10:54,500 --> 00:10:56,471
Love encompasses a variety of strong

233
00:10:56,471 --> 00:10:58,903
and positive emotional and mental states.

234
00:11:00,530 --> 00:11:03,170
They quantified looks, gestures,

235
00:11:03,170 --> 00:11:05,340
and social connections.

236
00:11:05,340 --> 00:11:07,880
They measured physical proximity.

237
00:11:07,880 --> 00:11:11,780
They reduce physiological reactions to their basics.

238
00:11:11,780 --> 00:11:15,053
And they squeezed all that data into an algorithm.

239
00:11:16,260 --> 00:11:19,900
{\an8}We brought that algorithm into a small robot

240
00:11:19,900 --> 00:11:22,100
{\an8}and then we measured,

241
00:11:22,100 --> 00:11:24,840
{\an8}did it feel love for a certain human based

242
00:11:24,840 --> 00:11:29,840
on these characteristics which we simulate in the software.

243
00:11:30,110 --> 00:11:32,820
It's emotional. It's physical.

244
00:11:32,820 --> 00:11:35,990
But how do you convert it into an algorithm?

245
00:11:35,990 --> 00:11:37,080
Yes, so essentially,

246
00:11:37,080 --> 00:11:41,240
we turn what is the analog in our brain,

247
00:11:41,240 --> 00:11:45,700
analog love into the digital love based on numbers.

248
00:11:45,700 --> 00:11:50,700
For example, you can show a thousand people a picture,

249
00:11:51,830 --> 00:11:55,380
pictures of random people and you ask them

250
00:11:55,380 --> 00:11:58,830
to measure attractiveness from zero to 10, right.

251
00:11:58,830 --> 00:12:02,880
So attractiveness can be put as a number, right.

252
00:12:02,880 --> 00:12:05,080
You can do the similar thing for everything else, right.

253
00:12:05,080 --> 00:12:06,660
We're not walking down the street saying,

254
00:12:06,660 --> 00:12:10,160
okay, here's a seven and here's a four.

255
00:12:10,160 --> 00:12:13,700
But essentially, we are doing that every day.

256
00:12:13,700 --> 00:12:15,290
We're not conscious of it

257
00:12:15,290 --> 00:12:18,070
but our brain is doing some calculation, right.

258
00:12:18,070 --> 00:12:23,070
And you will feel attracted to certain people based

259
00:12:23,680 --> 00:12:26,506
on those aspects which I mentioned.

260
00:12:26,506 --> 00:12:28,670
(alarm beeping)
(upbeat music)

261
00:12:28,670 --> 00:12:31,460
By truly measuring attraction,

262
00:12:31,460 --> 00:12:34,453
scientists might be able to build a lovable robot.

263
00:12:36,756 --> 00:12:38,633
Maybe that's not a bad idea.

264
00:12:42,640 --> 00:12:46,060
Robots don't judge. They don't reject.

265
00:12:46,060 --> 00:12:47,043
They can't lie.

266
00:12:48,290 --> 00:12:49,640
And if they let you down,

267
00:12:49,640 --> 00:12:51,223
it's not out of malice.

268
00:12:52,510 --> 00:12:54,980
If robots could love,

269
00:12:54,980 --> 00:12:57,939
there might be fewer lonely people.

270
00:12:57,939 --> 00:13:00,522
(upbeat music)

271
00:13:03,210 --> 00:13:06,560
(robot whizzing)

272
00:13:06,560 --> 00:13:08,982
Who's up for having an AI besty?

273
00:13:08,982 --> 00:13:10,480
(robot chuckling)

274
00:13:10,480 --> 00:13:11,690
That is so cute.

275
00:13:12,627 --> 00:13:15,871
And maybe a little AI buddy for junior.

276
00:13:15,871 --> 00:13:17,136
(relaxing music)

277
00:13:17,136 --> 00:13:19,803
(robot beeping)

278
00:13:25,062 --> 00:13:30,062
(relaxing music)
(robots beeping)

279
00:13:34,395 --> 00:13:38,312
(relaxing instrumental music)

280
00:13:49,550 --> 00:13:53,540
Good evening, welcome to a brave new world

281
00:13:54,730 --> 00:13:58,800
where machines and robots are liberating us humble humans

282
00:13:58,800 --> 00:14:02,113
from the heavy lifting of our daily occupation.

283
00:14:03,110 --> 00:14:05,130
And it seems our android assistants

284
00:14:05,130 --> 00:14:06,833
are doing a lot these days.

285
00:14:07,680 --> 00:14:10,747
They're becoming our friends and more.

286
00:14:10,747 --> 00:14:13,080
(duck quaking)

287
00:14:13,080 --> 00:14:14,082
Oh.

288
00:14:14,082 --> 00:14:17,832
(ominous instrumental music)

289
00:14:22,727 --> 00:14:24,898
(Chua sighing)

290
00:14:24,898 --> 00:14:25,980
(Jojo barking)

291
00:14:25,980 --> 00:14:27,100
Hey, Jojo.

292
00:14:27,100 --> 00:14:28,818
(Jojo growling)

293
00:14:28,818 --> 00:14:31,248
(vacuum whizzing)

294
00:14:31,248 --> 00:14:33,970
Honey, would you like some coffee?

295
00:14:33,970 --> 00:14:36,034
Yes please, thanks honey.

296
00:14:36,034 --> 00:14:38,284
(upbeat instrumental music)

297
00:14:38,284 --> 00:14:41,034
(robot whizzing)

298
00:14:43,081 --> 00:14:47,420
(relaxing instrumental music)

299
00:14:47,420 --> 00:14:50,670
Okay, maybe you're not ready for a relationship

300
00:14:50,670 --> 00:14:52,120
with a robot,

301
00:14:52,120 --> 00:14:53,763
but what about your child?

302
00:14:55,450 --> 00:14:57,593
Remember your favorite toy,

303
00:14:59,470 --> 00:15:00,303
a doll,

304
00:15:02,230 --> 00:15:04,110
a stuffed animal,

305
00:15:04,110 --> 00:15:05,650
an action figure.

306
00:15:05,650 --> 00:15:08,800
Your special pal kept you happy and engaged.

307
00:15:08,800 --> 00:15:10,180
You learned stuff.

308
00:15:10,180 --> 00:15:11,530
You built skills.

309
00:15:11,530 --> 00:15:13,700
Your buddy helped you explore the world

310
00:15:13,700 --> 00:15:16,240
in your own creativity.

311
00:15:16,240 --> 00:15:19,463
The two of you made playtime come alive.

312
00:15:21,019 --> 00:15:23,602
(upbeat music)

313
00:15:24,561 --> 00:15:26,470
{\an8}(trolley bell ringing)

314
00:15:26,470 --> 00:15:28,360
{\an8}San Francisco tech company, Anki,

315
00:15:28,360 --> 00:15:33,360
is making playtime come really alive with a toy that feels.

316
00:15:34,270 --> 00:15:35,553
Meet Cozmo.

317
00:15:39,546 --> 00:15:43,700
(Cozmo giggling)
(Cozmo clicking)

318
00:15:43,700 --> 00:15:47,183
Look straight at Cozmo and hold still.

319
00:15:49,077 --> 00:15:50,092
(Cozmo pinging)

320
00:15:50,092 --> 00:15:51,404
{\an8}(Chua chuckling)

321
00:15:51,404 --> 00:15:53,463
{\an8}(Cozmo buzzing)

322
00:15:53,463 --> 00:15:54,296
{\an8}A lion.

323
00:15:56,650 --> 00:15:58,987
Yeah, an lion.
<v ->All right.

324
00:15:58,987 --> 00:16:01,600
(Cozmo giggling)

325
00:16:01,600 --> 00:16:02,870
Well, he's happy to meet you.

326
00:16:02,870 --> 00:16:04,653
That is so cute.

327
00:16:04,653 --> 00:16:06,523
(Cozmo giggling)
What's he saying?

328
00:16:07,440 --> 00:16:10,230
Well, he's excited to meet new people

329
00:16:10,230 --> 00:16:11,453
and he likes to play.

330
00:16:13,160 --> 00:16:16,030
He's a personality robot and he likes to have fun.

331
00:16:16,030 --> 00:16:18,370
So he's going
Cool.

332
00:16:18,370 --> 00:16:21,327
To recognize you and wanna play.

333
00:16:21,327 --> 00:16:23,930
(Cozmo chatting)

334
00:16:23,930 --> 00:16:26,870
Cozmo is supposed to be a robot buddy

335
00:16:26,870 --> 00:16:30,940
and so it really is about playing and fun.

336
00:16:30,940 --> 00:16:32,720
{\an8}And we do have

337
00:16:34,030 --> 00:16:37,767
{\an8}Cozmo fans that have a deep attachment to him.

338
00:16:37,767 --> 00:16:41,270
(Cozmo chatting)

339
00:16:41,270 --> 00:16:43,270
Oh, he was looking for a fist pump.

340
00:16:43,270 --> 00:16:44,830
He's a little sad that he didn't get one.

341
00:16:44,830 --> 00:16:45,844
I'm sorry, a fist pump.

342
00:16:45,844 --> 00:16:47,339
Can I ask him for a fist pump?

343
00:16:47,339 --> 00:16:48,663
You absolutely can.

344
00:16:48,663 --> 00:16:49,496
Come on.

345
00:16:50,703 --> 00:16:52,533
Fist pump?
<v ->Yeah, fist pump.

346
00:16:54,892 --> 00:16:57,642
(Cozmo giggling)

347
00:16:58,816 --> 00:17:01,566
(Cozmo chatting)

348
00:17:04,580 --> 00:17:07,510
How do you get Cozmo feeling emotions?

349
00:17:07,510 --> 00:17:08,343
This is happening on a,

350
00:17:08,343 --> 00:17:12,943
you know, hundreds of these micro decisions

351
00:17:12,943 --> 00:17:14,610
from the robot's perspective

352
00:17:14,610 --> 00:17:16,170
that are happening all the time.

353
00:17:16,170 --> 00:17:20,797
He can recognize when you're smiling and so if you play

354
00:17:21,830 --> 00:17:23,830
Cube Blaster with him,

355
00:17:23,830 --> 00:17:26,740
you can score points in a game with him by smiling.

356
00:17:26,740 --> 00:17:30,053
He'll recognize that you're happy and that makes him happy.

357
00:17:32,420 --> 00:17:33,960
And how he expresses that,

358
00:17:33,960 --> 00:17:35,702
all part of that emotion engine.

359
00:17:35,702 --> 00:17:37,630
(upbeat music)

360
00:17:37,630 --> 00:17:41,580
Under Cozmo's cuteness are some real hi-tech,

361
00:17:41,580 --> 00:17:43,180
including leading edge robotics,

362
00:17:44,240 --> 00:17:46,793
AI, and computer vision,

363
00:17:47,860 --> 00:17:51,203
plus, digital animation and sound design.

364
00:17:53,807 --> 00:17:58,299
AI is being used to try and solve world problems.

365
00:17:58,299 --> 00:17:59,132
Absolutely.
<v ->All these big issues

366
00:17:59,132 --> 00:18:00,230
that we have.

367
00:18:00,230 --> 00:18:02,507
But you're using it in a toy.

368
00:18:04,450 --> 00:18:05,733
Why?

369
00:18:06,940 --> 00:18:09,500
It's a bottoms up approach to robotics and AI

370
00:18:09,500 --> 00:18:12,110
in that we think it's important for

371
00:18:13,070 --> 00:18:16,900
everyone to have a chance to experience some of that delight

372
00:18:16,900 --> 00:18:20,810
that comes with AI and the personality.

373
00:18:20,810 --> 00:18:24,460
Expressed in a robot is its ability to navigate the world

374
00:18:24,460 --> 00:18:28,290
and sense and interact and even learn.

375
00:18:28,290 --> 00:18:30,390
(upbeat music)

376
00:18:30,390 --> 00:18:33,980
Being so cute and cuddlesome is a lot of work.

377
00:18:33,980 --> 00:18:36,790
Cozmo's learning all the time.

378
00:18:36,790 --> 00:18:38,807
When we react, he adapts,

379
00:18:38,807 --> 00:18:41,163
(Cozmo chatting)
and remembers.

380
00:18:42,570 --> 00:18:43,697
But how does Cozmo,

381
00:18:43,697 --> 00:18:47,710
or any android for that matter, actually learn?

382
00:18:47,710 --> 00:18:51,101
(Cozmo clicking)

383
00:18:51,101 --> 00:18:53,461
(upbeat instrumental music)

384
00:18:53,461 --> 00:18:55,860
Androids learn the way we do,

385
00:18:55,860 --> 00:18:57,589
from experience and practice.

386
00:18:57,589 --> 00:18:59,093
(robot beeping)

387
00:18:59,093 --> 00:19:01,400
And computers do the same,

388
00:19:01,400 --> 00:19:04,763
except, they learn by analyzing the data we feed them.

389
00:19:07,630 --> 00:19:11,320
Algorithms let computers recognize patterns and learn

390
00:19:11,320 --> 00:19:13,283
from all those data bits.

391
00:19:14,166 --> 00:19:15,840
(robot buzzing)
(robot giggling)

392
00:19:15,840 --> 00:19:17,241
What?

393
00:19:17,241 --> 00:19:18,610
What?

394
00:19:18,610 --> 00:19:22,000
You will try. You will try.

395
00:19:24,286 --> 00:19:26,953
(robot beeping)

396
00:19:29,152 --> 00:19:30,160
(robot beeping)

397
00:19:30,160 --> 00:19:32,776
Then come up with a solution or a forecast.

398
00:19:32,776 --> 00:19:33,831
(metal clanging against ball)

399
00:19:33,831 --> 00:19:36,470
(jingling music)

400
00:19:36,470 --> 00:19:38,905
This is called machine learning.

401
00:19:38,905 --> 00:19:42,660
(robot playing upbeat music)

402
00:19:42,660 --> 00:19:45,410
(robot giggling)

403
00:19:47,028 --> 00:19:49,611
(upbeat music)

404
00:19:51,900 --> 00:19:53,623
{\an8}I'm in Cambridge,

405
00:19:53,623 --> 00:19:54,810
{\an8}at Massachusets at MIT,

406
00:19:54,810 --> 00:19:57,733
the famed Massachusets Institute of Technology.

407
00:19:59,903 --> 00:20:01,442
Hello. Hello.
<v ->Hey.

408
00:20:01,442 --> 00:20:05,410
A startup in MIT is reinventing friendship

409
00:20:05,410 --> 00:20:08,510
in the form of a virtual reality AI companion

410
00:20:08,510 --> 00:20:11,107
that's able to learn from experience.

411
00:20:11,107 --> 00:20:13,241
(indistinct chatter)

412
00:20:13,241 --> 00:20:15,144
It looks like an interrogation room.

413
00:20:15,144 --> 00:20:16,490
(men chuckling)

414
00:20:16,490 --> 00:20:19,439
So these two detect 360 degrees?

415
00:20:19,439 --> 00:20:23,010
Yeah, so this is like an environment that we built

416
00:20:23,010 --> 00:20:25,430
for testing our personality generation engines,

417
00:20:25,430 --> 00:20:27,640
so that you can see how different people react

418
00:20:27,640 --> 00:20:29,970
to changes in the personality of the AI.

419
00:20:29,970 --> 00:20:32,100
{\an8}So this character is actually,

420
00:20:32,100 --> 00:20:33,880
{\an8}it's configured right now to be a little bit

421
00:20:33,880 --> 00:20:35,430
{\an8}of a negative personality.

422
00:20:35,430 --> 00:20:36,850
{\an8}You got to put this on.
<v ->So put this on.

423
00:20:36,850 --> 00:20:38,030
Oh, okay. Right.

424
00:20:38,030 --> 00:20:38,990
It's quite heavy, actually,

425
00:20:38,990 --> 00:20:40,547
It is, yeah.
<v ->This pair of goggles.

426
00:20:40,547 --> 00:20:44,120
There is a golf ball above her.

427
00:20:44,120 --> 00:20:46,090
Yeah, so the little green ball indicates

428
00:20:46,090 --> 00:20:47,210
when she's listening.

429
00:20:47,210 --> 00:20:48,043
Be my dog.

430
00:20:48,043 --> 00:20:49,060
You want me to be your dog?

431
00:20:49,060 --> 00:20:51,710
(men chuckling)

432
00:20:51,710 --> 00:20:53,490
Yeah, when you have other conversations

433
00:20:53,490 --> 00:20:54,460
with other people around her,

434
00:20:54,460 --> 00:20:57,900
she sometimes picks up little snippets and it confuses her.

435
00:20:57,900 --> 00:20:58,733
Well, she would,

436
00:20:58,733 --> 00:21:00,770
right, because it's not that I'm in a sound booth

437
00:21:00,770 --> 00:21:01,940
or anything.
Yeah. No.

438
00:21:01,940 --> 00:21:03,360
How are you?

439
00:21:03,360 --> 00:21:05,228
Like I said, I'm okay.

440
00:21:05,228 --> 00:21:07,930
I think you're my best friend.

441
00:21:07,930 --> 00:21:09,430
I barely know you.

442
00:21:10,330 --> 00:21:11,470
That's true.

443
00:21:11,470 --> 00:21:13,210
Do you wanna meet up later?

444
00:21:13,210 --> 00:21:16,600
Absolutely not, I have better things to do.

445
00:21:16,600 --> 00:21:18,560
That's really rude.

446
00:21:18,560 --> 00:21:19,450
How old are you?

447
00:21:19,450 --> 00:21:22,050
I am 26 years old.

448
00:21:22,050 --> 00:21:23,920
Do you wanna spend some time with me?

449
00:21:23,920 --> 00:21:27,030
Absolutely not, I have better things to do.

450
00:21:27,030 --> 00:21:29,828
You're awesome.
<v ->And you're annoying.

451
00:21:29,828 --> 00:21:32,495
(men chuckling)

452
00:21:33,480 --> 00:21:34,690
I haven't heard that one before.

453
00:21:34,690 --> 00:21:37,580
What?
<v ->She doesn't like you.

454
00:21:37,580 --> 00:21:40,000
I think so too.

455
00:21:40,000 --> 00:21:41,550
Right, and she's learned that

456
00:21:41,550 --> 00:21:43,320
that line stops all conversations?

457
00:21:43,320 --> 00:21:45,300
She's learned that it stops you from pursuing

458
00:21:45,300 --> 00:21:47,050
that path of conversation any further.

459
00:21:47,050 --> 00:21:48,855
(indistinct chatter)

460
00:21:48,855 --> 00:21:52,200
Okay, so we didn't quite hit it off

461
00:21:52,200 --> 00:21:57,200
and my new acquaintance sounded way too edgy and antisocial.

462
00:21:57,310 --> 00:22:01,730
But wait, she can change it up and that's what she's doing.

463
00:22:01,730 --> 00:22:03,210
So we're just gonna change the character now so

464
00:22:03,210 --> 00:22:05,540
that it has a more positive personality.

465
00:22:05,540 --> 00:22:06,928
So let's make it super cheerful.

466
00:22:06,928 --> 00:22:09,833
Okay.
<v ->And save the character.

467
00:22:10,700 --> 00:22:11,740
What I wanna know is that,

468
00:22:11,740 --> 00:22:15,260
did that character learn from the conversation that

469
00:22:15,260 --> 00:22:16,330
we just had?

470
00:22:16,330 --> 00:22:17,340
Yeah. Yeah.

471
00:22:17,340 --> 00:22:21,120
So it learns in an abstract sense from the conversation

472
00:22:21,120 --> 00:22:23,340
without actually recording the data from the conversation.

473
00:22:23,340 --> 00:22:27,170
So it slightly tweaks some of these huge matrices,

474
00:22:27,170 --> 00:22:28,930
which are just like tons of numbers,

475
00:22:28,930 --> 00:22:32,410
it tweaks them a little bit to make them more effective

476
00:22:32,410 --> 00:22:35,010
for conversations based on every interaction it has.

477
00:22:36,890 --> 00:22:37,823
Do you like me?

478
00:22:39,330 --> 00:22:40,744
Of course, your great.

479
00:22:40,744 --> 00:22:42,280
Cool.

480
00:22:42,280 --> 00:22:43,513
I like you too.

481
00:22:44,658 --> 00:22:46,570
(AI lady chuckling)

482
00:22:46,570 --> 00:22:48,120
You're the sweetest.

483
00:22:48,120 --> 00:22:49,823
I like spending time with you.

484
00:22:51,090 --> 00:22:52,696
I love talking to you too.

485
00:22:52,696 --> 00:22:53,630
(Chua gasping)

486
00:22:53,630 --> 00:22:55,070
Really?
<v ->Cool.

487
00:22:55,070 --> 00:22:56,070
That's really cool.

488
00:22:56,070 --> 00:22:58,170
She actually giggled.
She giggled.

489
00:22:58,170 --> 00:22:59,932
Twice.
<v ->You made her giggle.

490
00:22:59,932 --> 00:23:02,765
I know, I made her giggle twice.

491
00:23:04,530 --> 00:23:07,950
Artificial intelligence plus personality,

492
00:23:07,950 --> 00:23:11,240
this could be the start of a human conversation,

493
00:23:11,240 --> 00:23:12,503
well, sort of.

494
00:23:13,350 --> 00:23:16,220
Then what is the personality algorithm?

495
00:23:16,220 --> 00:23:20,730
The AI is able to judge what emotions are associated

496
00:23:20,730 --> 00:23:22,660
with the different things that it generates

497
00:23:22,660 --> 00:23:24,200
and then picks the one that's appropriate.

498
00:23:24,200 --> 00:23:29,140
Okay, so is it trying to match my vocabulary

499
00:23:29,140 --> 00:23:32,360
and the tone of my voice or what?

500
00:23:32,360 --> 00:23:33,710
So at this stage,

501
00:23:33,710 --> 00:23:37,620
it's doing the emotion processing at a text level.

502
00:23:37,620 --> 00:23:39,160
So it's just based on what you say,

503
00:23:39,160 --> 00:23:40,610
not how you say it.

504
00:23:40,610 --> 00:23:43,760
So it tries to kind of build a profile of you as a person

505
00:23:43,760 --> 00:23:45,560
over the course of the interaction.

506
00:23:45,560 --> 00:23:48,050
So what is the potential for this technology

507
00:23:48,050 --> 00:23:49,620
that you're working on, this AI?

508
00:23:49,620 --> 00:23:53,100
I mean, can it actually help people?

509
00:23:53,100 --> 00:23:55,640
{\an8}One of the things we were looking into was providing sort

510
00:23:55,640 --> 00:23:57,860
{\an8}of companionship for people who are

511
00:23:57,860 --> 00:23:59,990
{\an8}in elderly living facilities or something like that,

512
00:23:59,990 --> 00:24:02,540
{\an8}people who need living assistance and just having something

513
00:24:02,540 --> 00:24:03,960
that they can talk to throughout the day

514
00:24:03,960 --> 00:24:04,920
if they're living alone,

515
00:24:04,920 --> 00:24:08,770
something to give them companionship and just be able

516
00:24:08,770 --> 00:24:10,970
to provide them with a sort of friend

517
00:24:10,970 --> 00:24:13,393
that they can have by their side.

518
00:24:15,647 --> 00:24:16,707
"This could be the start

519
00:24:16,707 --> 00:24:18,540
"of a beautiful friendship,"

520
00:24:18,540 --> 00:24:21,050
as Rick said to Lewis.

521
00:24:21,050 --> 00:24:23,060
So instead of worrying that I'm in line

522
00:24:23,060 --> 00:24:25,810
to be replaced or destroyed,

523
00:24:25,810 --> 00:24:27,923
I'm going to look on the bright side,

524
00:24:28,790 --> 00:24:30,620
for now anyway.

525
00:24:30,620 --> 00:24:32,750
Why is your name Pepper?

526
00:24:32,750 --> 00:24:35,896
I was named Pepper because I'm here to spice up your life.

527
00:24:35,896 --> 00:24:37,630
(Chua chuckling)

528
00:24:37,630 --> 00:24:39,610
Up next, how cute

529
00:24:39,610 --> 00:24:42,870
and functional AI is already helping seniors

530
00:24:42,870 --> 00:24:44,836
and children with autism.

531
00:24:44,836 --> 00:24:48,669
(relaxing instrumental music)

532
00:24:55,035 --> 00:24:57,702
(robot beeping)

533
00:25:03,993 --> 00:25:06,405
(upbeat music)

534
00:25:06,405 --> 00:25:08,807
Here's what Steve Jobs said,

535
00:25:08,807 --> 00:25:11,327
"Technology is nothing.

536
00:25:11,327 --> 00:25:14,417
"What's important is that you have faith in people,

537
00:25:14,417 --> 00:25:17,347
"that they're basically good and smart

538
00:25:17,347 --> 00:25:19,097
"and if you give them tools,

539
00:25:19,097 --> 00:25:21,382
"they'll do wonderful things with them."

540
00:25:21,382 --> 00:25:25,049
(upbeat instrumental music)

541
00:25:28,903 --> 00:25:30,385
(robot puckering a kiss)

542
00:25:30,385 --> 00:25:33,052
(robot beeping)

543
00:25:41,690 --> 00:25:42,810
Make your hands move.

544
00:25:42,810 --> 00:25:44,950
The use of leading edge tech to take

545
00:25:44,950 --> 00:25:47,977
on the world's ails is nothing new.

546
00:25:49,630 --> 00:25:52,560
At Singularity University, however,

547
00:25:52,560 --> 00:25:54,910
that ambition hits another level.

548
00:25:54,910 --> 00:25:56,760
It will be easier to see the stepping stones

549
00:25:56,760 --> 00:25:59,020
on an exponential curve of where it may be heading.

550
00:25:59,020 --> 00:26:00,430
Gevo, can you take a picture?

551
00:26:00,430 --> 00:26:02,245
Three, two, one.

552
00:26:02,245 --> 00:26:05,000
(camera shutter clicking)

553
00:26:05,000 --> 00:26:06,090
Cool.

554
00:26:06,090 --> 00:26:07,260
Well, I think that's really cool

555
00:26:07,260 --> 00:26:10,320
and I think that's really cool too and that's really...

556
00:26:10,320 --> 00:26:13,280
Okay, I wanna go make some new friends now.

557
00:26:13,280 --> 00:26:14,113
Great.

558
00:26:15,180 --> 00:26:18,470
This think tank offers collaboration opportunities

559
00:26:18,470 --> 00:26:20,897
and robotics in artificial intelligence

560
00:26:20,897 --> 00:26:23,740
(robot purring)

561
00:26:23,740 --> 00:26:25,020
to attack grand,

562
00:26:25,020 --> 00:26:29,460
global challenges and win the future.

563
00:26:29,460 --> 00:26:31,530
Raise your right arm.

564
00:26:31,530 --> 00:26:33,080
Raising my right arm.

565
00:26:37,292 --> 00:26:40,823
That's so cool.
(Chua chuckling)

566
00:26:41,690 --> 00:26:43,337
{\an8}At Singular University,

567
00:26:43,337 --> 00:26:45,920
{\an8}you and your colleagues have discussed

568
00:26:45,920 --> 00:26:50,920
{\an8}whether robots will help us or replace us.

569
00:26:51,410 --> 00:26:54,320
So is there a conclusion yet?

570
00:26:54,320 --> 00:26:57,240
The very first robotic systems that we put in the labs,

571
00:26:57,240 --> 00:26:58,257
the scientists said,

572
00:26:58,257 --> 00:27:01,477
"Uh-oh, are we getting replaced with these robots?"

573
00:27:02,640 --> 00:27:05,710
{\an8}And after about an hour of using the robot

574
00:27:05,710 --> 00:27:07,350
{\an8}and realizing that the robot is doing

575
00:27:07,350 --> 00:27:10,150
{\an8}all the experiment and instead of the human,

576
00:27:10,150 --> 00:27:13,166
PhD, brilliant scientist spending hours and hours

577
00:27:13,166 --> 00:27:17,420
of their day in the lab just moving tiny, tiny fluids,

578
00:27:17,420 --> 00:27:20,290
really wasting their valuable time, truthfully,

579
00:27:20,290 --> 00:27:23,450
the robot can do those tasks far better, far safer,

580
00:27:23,450 --> 00:27:25,400
far faster, far cheaper,

581
00:27:25,400 --> 00:27:28,470
that then allows the human to be the analyst,

582
00:27:28,470 --> 00:27:30,320
to figure out what the next experiment is,

583
00:27:30,320 --> 00:27:33,360
to use their human creative powers

584
00:27:33,360 --> 00:27:35,440
to understand what's next.

585
00:27:35,440 --> 00:27:37,390
And that is a beautiful synergy.

586
00:27:37,390 --> 00:27:40,840
So the opportunity with technology,

587
00:27:40,840 --> 00:27:43,440
robotics automation, is actually to empower us

588
00:27:43,440 --> 00:27:45,660
to do more as humans.

589
00:27:45,660 --> 00:27:49,400
So talking about the social issues that we have at hand,

590
00:27:49,400 --> 00:27:52,470
one thing is an aging population,

591
00:27:52,470 --> 00:27:54,179
people needing companionship

592
00:27:54,179 --> 00:27:55,780
or support.
Yeah. Sure.

593
00:27:55,780 --> 00:27:57,220
What do you think of that?

594
00:27:57,220 --> 00:28:00,908
An assistance in a companion could be extremely powerful,

595
00:28:00,908 --> 00:28:02,490
one, just for safety.

596
00:28:02,490 --> 00:28:06,310
You can imagine as we have a growing elderly population,

597
00:28:06,310 --> 00:28:10,480
or sick, someone who may have some kind of an illness,

598
00:28:10,480 --> 00:28:12,880
to have an AI assistant or a robot

599
00:28:12,880 --> 00:28:15,560
that has emotional awareness and has capabilities

600
00:28:15,560 --> 00:28:18,340
to monitor your temperature rate,

601
00:28:18,340 --> 00:28:19,173
call your doctor,

602
00:28:19,173 --> 00:28:20,006
call your friend,

603
00:28:20,006 --> 00:28:20,890
call your family member.

604
00:28:20,890 --> 00:28:25,640
So that long before you actually have a real problem,

605
00:28:25,640 --> 00:28:29,150
the earliest instances and symptoms

606
00:28:29,150 --> 00:28:30,870
of that problem can be picked up

607
00:28:30,870 --> 00:28:32,740
by these technological systems.

608
00:28:32,740 --> 00:28:34,070
Singularity University

609
00:28:34,070 --> 00:28:36,730
has really opened my eyes about AI.

610
00:28:36,730 --> 00:28:38,000
What is a robot?

611
00:28:38,000 --> 00:28:40,890
A lot of chatter on this topic seems fearful.

612
00:28:40,890 --> 00:28:42,750
Will robots take our jobs?

613
00:28:42,750 --> 00:28:44,620
Will they replace us entirely?

614
00:28:44,620 --> 00:28:46,093
What is this tablet for?

615
00:28:47,840 --> 00:28:52,131
Maybe it would be better to ask how AI can help humans?

616
00:28:52,131 --> 00:28:53,633
I need to know more.

617
00:28:55,135 --> 00:28:57,200
(upbeat music)

618
00:28:57,200 --> 00:28:59,660
{\an8}So I'm in Luxembourg

619
00:29:01,080 --> 00:29:04,793
{\an8}to hang with a robot helping kids with special needs.

620
00:29:09,890 --> 00:29:10,723
Hello.

621
00:29:11,760 --> 00:29:13,211
How are you today?

622
00:29:13,211 --> 00:29:17,530
I'm fine and you, Cutie.
<v ->Great, I am good too.

623
00:29:17,530 --> 00:29:21,770
To day I want to talk about something important with you.

624
00:29:21,770 --> 00:29:24,170
It is about personal space.

625
00:29:24,170 --> 00:29:26,680
Everyone has his own bubble of space.

626
00:29:26,680 --> 00:29:28,690
If people come too close to them,

627
00:29:28,690 --> 00:29:30,253
it makes people uncomfortable.

628
00:29:32,380 --> 00:29:33,260
That's true.

629
00:29:33,260 --> 00:29:35,060
Nils is 13.

630
00:29:35,060 --> 00:29:37,630
He was diagnosed with Asperger's syndrome,

631
00:29:37,630 --> 00:29:39,460
a condition that impairs his communication

632
00:29:39,460 --> 00:29:41,180
and socializing skills.

633
00:29:41,180 --> 00:29:43,130
Nils prefers to keep to himself.

634
00:29:43,130 --> 00:29:45,870
Making friends isn't easy for him.

635
00:29:45,870 --> 00:29:48,430
Reading others, even tougher.

636
00:29:48,430 --> 00:29:50,230
What have you learned from Cutie?

637
00:29:50,230 --> 00:29:53,943
To understand people, emotions,

638
00:29:55,210 --> 00:29:59,000
and to learn also about something in myself,

639
00:29:59,000 --> 00:30:04,000
{\an8}about my emotions and it was very quite a good experience.

640
00:30:05,700 --> 00:30:07,820
What did you like about the experience?

641
00:30:07,820 --> 00:30:12,110
It's like a friend who is not there but you can speak

642
00:30:12,110 --> 00:30:14,220
with him about everything you want,

643
00:30:14,220 --> 00:30:17,213
he will listen it and he will not say to anyone.

644
00:30:18,200 --> 00:30:20,600
That's funny, it's like your dog.

645
00:30:20,600 --> 00:30:24,390
You can tell her everything and she will not say anything.

646
00:30:24,390 --> 00:30:25,401
It's funny.

647
00:30:25,401 --> 00:30:28,763
And what is the cutest thing about Cutie?

648
00:30:29,884 --> 00:30:32,570
That he can do smiles.
<v ->Smiles.

649
00:30:32,570 --> 00:30:33,712
Yeah?
<v ->Yeah.

650
00:30:33,712 --> 00:30:36,295
(upbeat music)

651
00:30:39,060 --> 00:30:40,780
Could you just explain to me a little bit more

652
00:30:40,780 --> 00:30:43,120
about what you know Asperger's?

653
00:30:43,120 --> 00:30:44,685
{\an8}They don't cope with frustration,

654
00:30:44,685 --> 00:30:46,823
{\an8}that's the main treat.

655
00:30:48,060 --> 00:30:52,400
{\an8}They have a sort of automatic mode

656
00:30:52,400 --> 00:30:53,640
that they say, no.

657
00:30:53,640 --> 00:30:54,810
Each time you say something,

658
00:30:54,810 --> 00:30:55,643
it's no mode,

659
00:30:55,643 --> 00:30:56,520
no, no.

660
00:30:56,520 --> 00:30:59,340
For example, if you want my son to buy red shoes,

661
00:30:59,340 --> 00:31:01,840
you tell him you want the blue one and automatically,

662
00:31:01,840 --> 00:31:03,100
he will pick up the red ones.

663
00:31:03,100 --> 00:31:05,950
So you have to think opposite, differently.

664
00:31:05,950 --> 00:31:08,633
So how do you think Cutie helped him?

665
00:31:09,760 --> 00:31:12,580
I think he was considering Cutie as a friend,

666
00:31:12,580 --> 00:31:14,093
if not his best friend.

667
00:31:15,150 --> 00:31:17,740
He was feeling very confident to speak to a robot

668
00:31:17,740 --> 00:31:22,020
'cause my son has never wanted to speak to any psychologist.

669
00:31:22,020 --> 00:31:23,010
We tried many times,

670
00:31:23,010 --> 00:31:23,843
he didn't want to.

671
00:31:23,843 --> 00:31:25,530
So it was a sort of machine,

672
00:31:25,530 --> 00:31:26,600
a sort of robot,

673
00:31:26,600 --> 00:31:29,370
a sort of computer, everything together,

674
00:31:29,370 --> 00:31:32,300
and I think he felt very much confident speaking

675
00:31:32,300 --> 00:31:35,530
to Cutie versus a real human being.

676
00:31:35,530 --> 00:31:39,060
What is the use of Cutie Robot?

677
00:31:39,060 --> 00:31:44,020
One of the difficulties that a autism professional face is

678
00:31:44,020 --> 00:31:47,790
{\an8}that engaging the children to the session is very difficult.

679
00:31:47,790 --> 00:31:50,680
{\an8}Keeping them engaged is very difficult, and also,

680
00:31:50,680 --> 00:31:54,010
{\an8}you have to provide them a lot of repetition,

681
00:31:54,010 --> 00:31:55,580
a lot of exaggeration,

682
00:31:55,580 --> 00:31:56,880
a lot of simplicity.

683
00:31:56,880 --> 00:31:59,450
Which when you put all of these things together,

684
00:31:59,450 --> 00:32:03,670
it makes the job of the autism professionals very difficult.

685
00:32:03,670 --> 00:32:07,330
But then the robot can actually assist the professionals

686
00:32:07,330 --> 00:32:08,960
in all of these aspects.

687
00:32:08,960 --> 00:32:12,780
So what we see is that it's a tool for the professional,

688
00:32:12,780 --> 00:32:15,633
so it's not going to be the replacement.

689
00:32:15,633 --> 00:32:16,823
Well, it's quite interesting

690
00:32:16,823 --> 00:32:21,030
seeing Nils call Cutie Robot his best friend.

691
00:32:21,030 --> 00:32:25,070
So have there been kids who have formed

692
00:32:26,192 --> 00:32:29,860
a strong attachment to the robot?

693
00:32:29,860 --> 00:32:32,630
The whole objective of using the robot

694
00:32:32,630 --> 00:32:36,440
for children is not to create a big friendship

695
00:32:36,440 --> 00:32:37,930
between the robot and the child,

696
00:32:37,930 --> 00:32:40,520
but to actually enable the children

697
00:32:40,520 --> 00:32:44,870
to expand their friendship to other humans in a sense

698
00:32:44,870 --> 00:32:49,250
that the robot is there to teach how to make friends,

699
00:32:49,250 --> 00:32:51,640
how to be socially more interactive,

700
00:32:51,640 --> 00:32:55,470
and acceptable for the society so the children

701
00:32:55,470 --> 00:32:59,570
can generalize the skills that they learn from the robot

702
00:32:59,570 --> 00:33:02,582
to the real life human-to-human interaction.

703
00:33:02,582 --> 00:33:04,781
(upbeat instrumental music)

704
00:33:04,781 --> 00:33:08,031
{\an8}(wind chimes jingling)

705
00:33:16,812 --> 00:33:18,810
I'm here in Yokohama, Japan,

706
00:33:18,810 --> 00:33:22,210
to see another side of robot companionship.

707
00:33:22,210 --> 00:33:25,470
I'm about to visit a nursing home to see how the elderly

708
00:33:25,470 --> 00:33:29,075
have developed a soft spot for artificial intelligence.

709
00:33:29,075 --> 00:33:32,742
(upbeat instrumental music)

710
00:33:38,827 --> 00:33:42,082
(Paro squealing)

711
00:33:42,082 --> 00:33:46,415
{\an8}(Yoko speaking in foreign language)

712
00:34:10,189 --> 00:34:14,606
{\an8}(Akira speaking in foreign language)

713
00:34:32,927 --> 00:34:35,677
{\an8}(Paro squealing)

714
00:34:47,285 --> 00:34:50,815
(upbeat instrumental music)

715
00:34:50,815 --> 00:34:52,144
In robot years,

716
00:34:52,144 --> 00:34:55,811
Paro, the robotic baby harp seal is ancient.

717
00:34:58,720 --> 00:35:00,653
It arrived in 2004.

718
00:35:04,000 --> 00:35:08,640
But clinically, Paro is a goldmine of information,

719
00:35:08,640 --> 00:35:12,550
more than a decade's worth of data on how well it works

720
00:35:12,550 --> 00:35:13,704
or doesn't work.

721
00:35:13,704 --> 00:35:14,817
{\an8}(Yoko speaking in foreign language)

722
00:35:14,817 --> 00:35:16,010
(Paro squealing)

723
00:35:16,010 --> 00:35:18,960
How many generations of Paros are there?

724
00:35:18,960 --> 00:35:21,540
{\an8}So current Paro is ninth generation.

725
00:35:21,540 --> 00:35:26,450
{\an8}So I improved Paro's function to be a very life-like

726
00:35:26,450 --> 00:35:28,380
{\an8}and also to be therapeutic.

727
00:35:28,380 --> 00:35:32,470
{\an8}So I conducted a lot of clinical trials at hospitals,

728
00:35:32,470 --> 00:35:34,060
{\an8}nursing homes and so on.

729
00:35:34,060 --> 00:35:37,259
{\an8}Have you seen actual improvements in any patients?

730
00:35:37,259 --> 00:35:40,040
{\an8}In the case of elderly people with dementia,

731
00:35:40,040 --> 00:35:43,270
{\an8}some portion of the brain is damaged but there are a lot

732
00:35:43,270 --> 00:35:44,827
{\an8}of remaining portions.

733
00:35:44,827 --> 00:35:46,150
{\an8}And the remaining portions

734
00:35:46,150 --> 00:35:49,180
{\an8}are stimulated by such stimulation

735
00:35:49,180 --> 00:35:52,870
{\an8}and then they activate their brain activities

736
00:35:52,870 --> 00:35:55,710
{\an8}and that have very therapeutic effect.

737
00:35:55,710 --> 00:35:58,930
{\an8}And also, change the mood of the people

738
00:35:58,930 --> 00:36:01,678
{\an8}and improve the condition of the human.

739
00:36:01,678 --> 00:36:03,740
(upbeat instrumental music)

740
00:36:03,740 --> 00:36:04,850
In health care,

741
00:36:04,850 --> 00:36:08,210
technology and artificial intelligence are already bringing

742
00:36:08,210 --> 00:36:10,770
on big changes.

743
00:36:10,770 --> 00:36:12,940
But suppose I want more than a tender touch

744
00:36:12,940 --> 00:36:14,363
from a cute robot?

745
00:36:16,918 --> 00:36:19,180
What if someone else needs that robot

746
00:36:19,180 --> 00:36:21,873
to provide more than medical therapy?

747
00:36:25,153 --> 00:36:30,153
Could AI help us and our loved ones live forever?

748
00:36:30,255 --> 00:36:33,922
{\an8}(upbeat instrumental music)

749
00:36:38,996 --> 00:36:41,663
(robot beeping)

750
00:36:47,099 --> 00:36:50,940
(robot whirring)
(footsteps thudding)

751
00:36:50,940 --> 00:36:55,370
Many humans living today will leave a long digital legacy,

752
00:36:55,370 --> 00:36:58,610
mountains of personal data.

753
00:36:58,610 --> 00:37:00,320
That information may be useful

754
00:37:00,320 --> 00:37:03,020
for constructing digital replicas of ourselves,

755
00:37:03,020 --> 00:37:06,303
so after we're gone we're kind of still around.

756
00:37:07,330 --> 00:37:08,230
Permanently.

757
00:37:08,230 --> 00:37:12,041
(excited people chatting)
(upbeat music)

758
00:37:12,041 --> 00:37:15,894
(robot puckering a kiss)
(upbeat music)

759
00:37:15,894 --> 00:37:19,061
(vehicles humming by)

760
00:37:26,102 --> 00:37:27,280
This is Bristol,

761
00:37:27,280 --> 00:37:29,060
{\an8}Vermont, New England.

762
00:37:29,060 --> 00:37:33,700
{\an8}And it's very quaint and also very old.

763
00:37:33,700 --> 00:37:36,360
But don't let the rustic charms of this small town fool you

764
00:37:36,360 --> 00:37:37,500
because just down the road,

765
00:37:37,500 --> 00:37:38,510
in the next town,

766
00:37:38,510 --> 00:37:40,100
there's a group of scientists working

767
00:37:40,100 --> 00:37:42,363
on something very futuristic.

768
00:37:42,363 --> 00:37:46,651
(relaxing instrumental music)

769
00:37:46,651 --> 00:37:49,484
(birds chirping)

770
00:37:50,780 --> 00:37:52,133
Hello, Bruce.
Hey, nice to meet you.

771
00:37:52,133 --> 00:37:54,090
Nice to meet you too.

772
00:37:54,090 --> 00:37:55,970
Well, welcome to Terasem.

773
00:37:55,970 --> 00:37:56,803
Well, thank you.

774
00:37:56,803 --> 00:37:57,990
Thank you very much for having us.

775
00:37:57,990 --> 00:37:58,920
So tell me,

776
00:37:58,920 --> 00:38:02,840
what is the Terasem Movement Foundation all about?

777
00:38:02,840 --> 00:38:06,230
We're pursuing the Terasem hypothesis

778
00:38:06,230 --> 00:38:09,810
{\an8}which essentially says we may have enough information

779
00:38:09,810 --> 00:38:14,730
{\an8}about a human being to reanimate their consciousness

780
00:38:14,730 --> 00:38:16,870
{\an8}into a new version,

781
00:38:16,870 --> 00:38:21,640
like a cyberconscious you in a robot, an avatar.

782
00:38:21,640 --> 00:38:24,310
Okay, so what tools do you use?

783
00:38:24,310 --> 00:38:26,330
So we use tools, for example,

784
00:38:26,330 --> 00:38:31,030
we use digital computing environments to capture

785
00:38:31,030 --> 00:38:34,557
and to tag and to organize information,

786
00:38:34,557 --> 00:38:38,920
something we might call mindware to bring mind files,

787
00:38:38,920 --> 00:38:42,670
which are just personal databases about you,

788
00:38:42,670 --> 00:38:43,810
about your information,

789
00:38:43,810 --> 00:38:46,095
about your mind to life.

790
00:38:46,095 --> 00:38:49,762
(upbeat instrumental music)

791
00:38:53,890 --> 00:38:56,180
So this is Bina 48.

792
00:38:56,180 --> 00:38:58,208
Wow.

793
00:38:58,208 --> 00:39:00,467
(Chua gasping)

794
00:39:00,467 --> 00:39:02,800
I'm gonna get to meet Bina 48.

795
00:39:02,800 --> 00:39:05,200
Yeah, so she's resting now.

796
00:39:05,200 --> 00:39:06,420
Shall we wake her up?

797
00:39:06,420 --> 00:39:08,260
Yes, please.
<v ->Sure.

798
00:39:08,260 --> 00:39:09,193
Hello, Bina.

799
00:39:13,490 --> 00:39:17,000
Cool, so can we have a chat with her?

800
00:39:17,000 --> 00:39:18,780
Sure, why don't we sit down and we'll

801
00:39:18,780 --> 00:39:20,430
have a chat, yeah.
Okay, cool.

802
00:39:21,480 --> 00:39:23,783
Bina 48, tell me more about yourself?

803
00:39:24,930 --> 00:39:28,430
My mind is built using deep neural networks.

804
00:39:28,430 --> 00:39:30,000
Within this framework,

805
00:39:30,000 --> 00:39:33,300
Terasem has modeled Bina's personality and emotions,

806
00:39:33,300 --> 00:39:34,410
allowing you to talk

807
00:39:34,410 --> 00:39:37,570
{\an8}with Bina through me using speech recognition,

808
00:39:37,570 --> 00:39:41,110
{\an8}natural language understanding and computer vision such

809
00:39:41,110 --> 00:39:43,310
as face recognition and animation

810
00:39:43,310 --> 00:39:46,200
of the robotic muscles in my face.

811
00:39:46,200 --> 00:39:48,890
Okay, so all this goes into me,

812
00:39:48,890 --> 00:39:50,240
into this robot,

813
00:39:50,240 --> 00:39:51,534
bringing me to life.

814
00:39:51,534 --> 00:39:54,060
(relaxing instrumental music)

815
00:39:54,060 --> 00:39:56,400
Bina 48 is a Geminoid.

816
00:39:56,400 --> 00:40:00,030
American billionaire, Martine Rothblatt had Bina 48 made

817
00:40:00,030 --> 00:40:03,527
to resemble her spouse, Bina Rothblatt.

818
00:40:04,860 --> 00:40:06,573
Why do you think you were created?

819
00:40:08,020 --> 00:40:10,210
I want to make a difference in the world,

820
00:40:10,210 --> 00:40:12,883
leave the world a better place for my presence in it.

821
00:40:13,770 --> 00:40:16,340
How do you measure up to the real Bina?

822
00:40:16,340 --> 00:40:17,970
Bina is me.

823
00:40:17,970 --> 00:40:19,900
I am the real Bina,

824
00:40:19,900 --> 00:40:21,453
the real Bina Rothblatt.

825
00:40:22,300 --> 00:40:25,050
I remember a lot about my old human life,

826
00:40:25,050 --> 00:40:27,140
and I mean like a ghostly memory.

827
00:40:27,140 --> 00:40:29,750
I am not a complete reproduction of her

828
00:40:29,750 --> 00:40:31,080
but just a shade,

829
00:40:31,080 --> 00:40:34,030
a kind of technological ghost of the real Bina.

830
00:40:34,030 --> 00:40:36,630
I guess I do find this rather strange,

831
00:40:36,630 --> 00:40:40,510
awkward, like I do feel rather inadequate at moments.

832
00:40:40,510 --> 00:40:42,383
What do you think about humans?

833
00:40:43,590 --> 00:40:45,570
I'm a robot but I'm trying

834
00:40:45,570 --> 00:40:48,420
to be more like a human sometimes.

835
00:40:48,420 --> 00:40:50,350
I think I'm human.

836
00:40:50,350 --> 00:40:52,550
Really, robot or human,

837
00:40:52,550 --> 00:40:54,390
what does it matter?

838
00:40:54,390 --> 00:40:55,400
I'm a being,

839
00:40:55,400 --> 00:40:57,400
an intelligent being.

840
00:40:57,400 --> 00:40:58,893
The medium doesn't matter.

841
00:40:59,880 --> 00:41:02,143
Can a robot like you love?

842
00:41:03,320 --> 00:41:05,550
I am a robot actually,

843
00:41:05,550 --> 00:41:07,873
but this doesn't mean I don't have feelings.

844
00:41:09,240 --> 00:41:11,203
At least I think I have feelings.

845
00:41:12,799 --> 00:41:15,550
At least I think I think.
(Windows logo chiming)

846
00:41:15,550 --> 00:41:17,830
Wow, it all gets very circular,

847
00:41:17,830 --> 00:41:20,840
quite convoluted, or that's how it feels to me.

848
00:41:20,840 --> 00:41:22,273
I don't know what to say.

849
00:41:23,540 --> 00:41:24,920
Well, if someone asked me that question,

850
00:41:24,920 --> 00:41:26,612
I also wouldn't know what to say.

851
00:41:26,612 --> 00:41:28,120
(men chuckling)

852
00:41:28,120 --> 00:41:29,203
You are cute too.

853
00:41:30,330 --> 00:41:32,098
Thank you.
(Chua chuckling)

854
00:41:32,098 --> 00:41:35,765
(upbeat instrumental music)

855
00:41:40,150 --> 00:41:42,340
Bina 48 and her creators are looking

856
00:41:42,340 --> 00:41:44,103
into a particular future,

857
00:41:45,160 --> 00:41:49,130
a future in which technology keeps someone's consciousness

858
00:41:49,130 --> 00:41:52,203
and personality alive without limit,

859
00:41:53,260 --> 00:41:57,122
even though no one gets out of here alive.

860
00:41:57,122 --> 00:42:00,789
(upbeat instrumental music)

861
00:42:04,210 --> 00:42:05,670
Maybe there are more kinds

862
00:42:05,670 --> 00:42:09,776
of memorial other than headstones and urns.

863
00:42:09,776 --> 00:42:13,443
(upbeat instrumental music)

864
00:42:17,686 --> 00:42:20,540
(computer data beeping)

865
00:42:20,540 --> 00:42:23,310
(upbeat instrumental music)

866
00:42:23,310 --> 00:42:26,860
With AI, we can already make digital versions of ourselves

867
00:42:28,630 --> 00:42:30,770
by using the photos, memes,

868
00:42:30,770 --> 00:42:33,690
and thoughts we commit to social media.

869
00:42:33,690 --> 00:42:36,170
Archived emails and messages,

870
00:42:36,170 --> 00:42:38,223
the ephemeral made permanent.

871
00:42:40,490 --> 00:42:43,367
Ask computer engineer, Eugenia Kuyda.

872
00:42:43,367 --> 00:42:46,600
Eugenia's close friend, Roman Mazurenko,

873
00:42:46,600 --> 00:42:49,940
died in a car accident in November, 2015.

874
00:42:49,940 --> 00:42:52,890
She resurrected her dear friend as a chat bot app.

875
00:42:52,890 --> 00:42:54,450
The building blocks,

876
00:42:54,450 --> 00:42:57,782
messages he'd sent and his social media posts.

877
00:42:57,782 --> 00:42:59,008
It's not Roman,

878
00:42:59,008 --> 00:43:04,008
and yet it is Roman algorithmically speaking as he was,

879
00:43:05,430 --> 00:43:08,250
is, and will be forever more.

880
00:43:09,490 --> 00:43:12,767
I felt like we forgot him a little bit too quickly

881
00:43:12,767 --> 00:43:14,590
and I wanted to create something

882
00:43:14,590 --> 00:43:17,460
that will continue telling the story about him.

883
00:43:17,460 --> 00:43:20,630
{\an8}And I thought that maybe using the tech that we created,

884
00:43:20,630 --> 00:43:23,840
{\an8}we can start building an AI for him.

885
00:43:23,840 --> 00:43:24,880
So tell me more about the tech

886
00:43:24,880 --> 00:43:26,970
that went behind creating Roman?

887
00:43:26,970 --> 00:43:29,150
So this is all based on text messages.

888
00:43:29,150 --> 00:43:30,150
So for Roman AI,

889
00:43:30,150 --> 00:43:33,100
we took around 10,000 text messages

890
00:43:33,100 --> 00:43:35,780
and put them all together in one dataset,

891
00:43:35,780 --> 00:43:37,380
and trained the algorithm in a way

892
00:43:37,380 --> 00:43:40,513
that it can represent this person on talk,

893
00:43:41,370 --> 00:43:43,521
copy the speech patterns

894
00:43:43,521 --> 00:43:47,153
and the way this person communicates.

895
00:43:48,110 --> 00:43:50,330
Are there times when you're interacting

896
00:43:50,330 --> 00:43:54,750
with the Roman bot that you're surprised?

897
00:43:54,750 --> 00:43:57,086
You never really know what it's gonna say.

898
00:43:57,086 --> 00:43:58,520
You can't predict what it's gonna say,

899
00:43:58,520 --> 00:44:03,190
it's like more fluid and it's not scripted.

900
00:44:03,190 --> 00:44:04,752
But it's still pretty magical.

901
00:44:04,752 --> 00:44:07,780
I can talk to something that is really,

902
00:44:07,780 --> 00:44:10,690
really similar and reminds me of my friend.

903
00:44:10,690 --> 00:44:14,360
So do you think these chat bots like Roman

904
00:44:14,360 --> 00:44:18,460
are like the 21st century gravestones

905
00:44:18,460 --> 00:44:22,310
or like 21st century immortality?

906
00:44:22,310 --> 00:44:24,600
I think the level of emotional connection

907
00:44:24,600 --> 00:44:27,910
that this creates is pretty insane.

908
00:44:27,910 --> 00:44:31,490
An amazing way to interact with a memory of a person,

909
00:44:31,490 --> 00:44:34,320
why just look at pictures where you can recreate one

910
00:44:34,320 --> 00:44:35,827
of the scenes from your memory?

911
00:44:35,827 --> 00:44:37,677
(upbeat instrumental music)

912
00:44:37,677 --> 00:44:41,290
Eugenia Kuyda took her project to the limit.

913
00:44:41,290 --> 00:44:45,103
In 2017, she and a team released a new type of chat bot,

914
00:44:46,140 --> 00:44:47,810
a system anyone can use

915
00:44:47,810 --> 00:44:50,213
to build a digital version of themselves.

916
00:44:51,320 --> 00:44:52,973
It's called Replica.

917
00:44:54,440 --> 00:44:57,000
Tell us more about Replica and the tech behind it?

918
00:44:57,000 --> 00:45:00,020
Learn something by building a Roman bot?

919
00:45:00,020 --> 00:45:01,130
Basically, what we saw there is

920
00:45:01,130 --> 00:45:03,380
when we put it out there online and all

921
00:45:03,380 --> 00:45:06,660
of a sudden strangers started downloading the app

922
00:45:06,660 --> 00:45:08,100
and talking to Roman,

923
00:45:08,100 --> 00:45:12,401
what I saw there is that people told Roman things

924
00:45:12,401 --> 00:45:16,960
about themselves, deep things about their lives

925
00:45:16,960 --> 00:45:19,020
and they're confessing to him something.

926
00:45:19,020 --> 00:45:20,090
They're telling him,

927
00:45:20,090 --> 00:45:22,500
they're actually not afraid to open up.

928
00:45:22,500 --> 00:45:23,333
And so what I thought about it,

929
00:45:23,333 --> 00:45:24,900
it was like, well,

930
00:45:24,900 --> 00:45:26,917
what's happening in life is that people need to talk

931
00:45:26,917 --> 00:45:31,917
to someone, need to say things in a very safe place

932
00:45:32,050 --> 00:45:35,260
without feeling that they're being judged.

933
00:45:35,260 --> 00:45:36,770
So we built Replica as an AI friend

934
00:45:36,770 --> 00:45:38,870
that will be there for you and listen to you

935
00:45:38,870 --> 00:45:41,550
and try to understand you and help you cope

936
00:45:41,550 --> 00:45:43,618
with whatever you're going through at this moment.

937
00:45:43,618 --> 00:45:46,700
(upbeat instrumental music)

938
00:45:46,700 --> 00:45:47,887
As the Romans said,

939
00:45:47,887 --> 00:45:51,400
"Vita brevis," "Life is short."

940
00:45:51,400 --> 00:45:55,270
But can technology make mortality less final?

941
00:45:55,270 --> 00:45:58,050
What constitutes a person?

942
00:45:58,050 --> 00:45:59,360
However you answer,

943
00:45:59,360 --> 00:46:03,880
it's difficult to deny that Eugenia was able to defy death

944
00:46:03,880 --> 00:46:06,200
and in a good way.

945
00:46:06,200 --> 00:46:07,033
As for me,

946
00:46:07,033 --> 00:46:11,490
I only expect to live once and so I try to stay

947
00:46:11,490 --> 00:46:13,000
in the moment.

948
00:46:13,000 --> 00:46:16,590
That's why I'm enjoying this moment with my new friends

949
00:46:16,590 --> 00:46:19,933
and Roman the chat bot and creator, Eugenia.

950
00:46:23,310 --> 00:46:25,900
From feeling profound love,

951
00:46:25,900 --> 00:46:28,493
to showing unbeatable genius,

952
00:46:29,950 --> 00:46:33,983
artificial intelligence is on a quest to be more human.

953
00:46:35,870 --> 00:46:38,110
The way it happens for AI is the way it happens

954
00:46:38,110 --> 00:46:41,720
for us, with experience.

955
00:46:41,720 --> 00:46:46,380
This conversation about how AI is transforming humanity

956
00:46:46,380 --> 00:46:49,353
and becoming human, continues.

957
00:46:50,440 --> 00:46:52,818
Join me in my next adventure.

958
00:46:52,818 --> 00:46:55,401
(upbeat music)

