1
00:00:00,948 --> 00:00:03,783
Hi. I'm Refik. I'm a media artist.

2
00:00:03,783 --> 00:00:05,618
I use data as a pigment

3
00:00:05,618 --> 00:00:07,888
and paint with a thinking brush

4
00:00:07,888 --> 00:00:10,747
that is assisted
by artificial intelligence.

5
00:00:11,364 --> 00:00:14,250
Using architectural spaces as canvases,

6
00:00:14,250 --> 00:00:15,818
I collaborate with machines

7
00:00:15,818 --> 00:00:18,737
to make buildings dream and hallucinate.

8
00:00:19,037 --> 00:00:21,680
You may be wondering,
what does all this mean?

9
00:00:21,680 --> 00:00:24,632
So let me please take you
into my work and my world.

10
00:00:24,632 --> 00:00:28,453
I witnessed the power of imagination

11
00:00:28,453 --> 00:00:30,110
when I was eight years old

12
00:00:30,110 --> 00:00:32,279
as a child growing up in Istanbul.

13
00:00:32,279 --> 00:00:35,583
One day, my mom brought home
a videocassette

14
00:00:35,583 --> 00:00:38,552
of the science fiction movie
"Blade Runner."

15
00:00:38,552 --> 00:00:41,345
I clearly remember being mesmerized

16
00:00:41,345 --> 00:00:46,277
by the stunning architectural vision
of the future of Los Angeles,

17
00:00:46,277 --> 00:00:48,697
a place that I had never seen before.

18
00:00:48,913 --> 00:00:52,368
That vision became a kind of a staple

19
00:00:52,368 --> 00:00:54,376
of my daydreams.

20
00:00:54,376 --> 00:00:57,011
When I arrived in LA in 2012

21
00:00:57,011 --> 00:00:59,914
for a graduate program
in design media arts,

22
00:00:59,914 --> 00:01:02,166
I rented a car and drove downtown

23
00:01:02,166 --> 00:01:04,675
to see that wonderful world
of the near future.

24
00:01:05,178 --> 00:01:08,431
I remember a specific line
that kept playing

25
00:01:08,431 --> 00:01:10,567
over and over in my head:

26
00:01:10,567 --> 00:01:14,192
the scene where the android Rachael

27
00:01:14,192 --> 00:01:16,744
realizes that her memories
are actually not hers,

28
00:01:16,744 --> 00:01:21,198
and when Deckard tells her
they are someone else's memories.

29
00:01:21,198 --> 00:01:22,540
Since that moment,

30
00:01:22,540 --> 00:01:24,109
one of my inspirations

31
00:01:24,109 --> 00:01:25,828
has been this question.

32
00:01:25,828 --> 00:01:30,247
What can a mission do
with someone else's memories?

33
00:01:30,247 --> 00:01:32,973
Or, to say that in another way,

34
00:01:32,973 --> 00:01:37,010
what does it mean to be an AI
in the 21st century?

35
00:01:37,748 --> 00:01:41,133
Any android or AI mission
is only intelligent

36
00:01:41,133 --> 00:01:43,586
as long as we collaborate with it.

37
00:01:43,586 --> 00:01:46,354
It can construct things
that human intelligence

38
00:01:46,354 --> 00:01:48,323
intends to produce,

39
00:01:48,323 --> 00:01:51,325
but does not have the capacity to do so.

40
00:01:51,325 --> 00:01:54,029
Think about your activities
and social networks.

41
00:01:54,029 --> 00:01:59,039
For example, they get smarter
the more you interact with them.

42
00:01:59,039 --> 00:02:03,593
If machines can learn or process memories,

43
00:02:03,593 --> 00:02:05,362
can they also dream?

44
00:02:05,362 --> 00:02:06,768
Hallucinate?

45
00:02:06,768 --> 00:02:09,438
Involuntarily remember,

46
00:02:09,438 --> 00:02:10,866
or make connections

47
00:02:10,866 --> 00:02:13,501
between multiple people streams?

48
00:02:13,501 --> 00:02:16,995
Does being an AI in the 21st century

49
00:02:16,995 --> 00:02:20,648
simply mean not forgetting anything?

50
00:02:20,648 --> 00:02:24,142
And, if so,

51
00:02:24,142 --> 00:02:25,410
isn't it the most revolutionary thing

52
00:02:25,410 --> 00:02:28,145
that we have experienced

53
00:02:28,145 --> 00:02:31,665
in our centuries-long effort
to capture history across media?

54
00:02:31,665 --> 00:02:33,500
In other words,

55
00:02:33,500 --> 00:02:36,977
how far have we come
since we discussed "Blade Runner"?

56
00:02:36,977 --> 00:02:40,487
So I established my studio in 2014,

57
00:02:40,487 --> 00:02:42,805
and invited architects,

58
00:02:42,805 --> 00:02:46,229
computer and data scientists,
neuroscientists, musicians

59
00:02:46,229 --> 00:02:48,075
and even storytellers to join me
in realizing my dreams.

60
00:02:48,075 --> 00:02:54,222
Can data become a pigment?

61
00:02:54,222 --> 00:02:57,608
This was the very first question we asked

62
00:02:57,608 --> 00:02:59,243
when starting our journey

63
00:02:59,243 --> 00:03:01,536
to embed media arts into architecture,

64
00:03:01,536 --> 00:03:04,838
to collide virtual and physical worlds.

65
00:03:04,838 --> 00:03:06,907
So we began to imagine

66
00:03:06,907 --> 00:03:08,208
what I would call

67
00:03:08,208 --> 00:03:10,294
the politics of data.

68
00:03:10,486 --> 00:03:13,053
One of our first projects,
Virtual Depictions,

69
00:03:13,053 --> 00:03:15,088
was a public data sculpture piece

70
00:03:15,088 --> 00:03:17,958
commissioned by the city of San Francisco.

71
00:03:17,958 --> 00:03:20,042
The work invites the audience

72
00:03:20,042 --> 00:03:23,487
to be part of a spectacular
aesthetic experience

73
00:03:23,487 --> 00:03:25,087
in a living urban space

74
00:03:25,087 --> 00:03:30,832
by depicting a ?? network
of connections of the city itself.

75
00:03:30,832 --> 00:03:35,086
It also stands as a reminder
of how invisible data

76
00:03:35,086 --> 00:03:36,739
from our everyday lives,

77
00:03:36,739 --> 00:03:39,574
like the ?? feeds
that are represented here,

78
00:03:39,574 --> 00:03:42,066
can be made visible

79
00:03:42,066 --> 00:03:45,335
and transformed into sensory knowledge

80
00:03:45,335 --> 00:03:48,305
that can be experienced collectively.

81
00:03:48,839 --> 00:03:52,699
In fact, data can only become knowledge

82
00:03:52,699 --> 00:03:54,267
when it's experienced,

83
00:03:54,267 --> 00:03:56,387
and what is knowledge and experience

84
00:03:56,387 --> 00:03:57,920
can take many forms.

85
00:03:57,920 --> 00:03:59,727
When exploring such connections

86
00:03:59,727 --> 00:04:02,832
through the vast potential
of machine intelligence,

87
00:04:03,350 --> 00:04:05,618
we also pondered

88
00:04:05,618 --> 00:04:09,539
the connection between human senses

89
00:04:09,539 --> 00:04:13,432
and the machines' capacity
for simulating nature.

90
00:04:13,432 --> 00:04:18,136
These inquiries began while working
on wind data paintings.

91
00:04:18,136 --> 00:04:20,490
They took the shape of visualized plans

92
00:04:20,490 --> 00:04:23,182
based on hidden data sets

93
00:04:23,182 --> 00:04:25,884
that we collected from wind sensors.

94
00:04:25,884 --> 00:04:28,602
We then used generative algorithms

95
00:04:28,602 --> 00:04:33,031
to transform wind speed,
gust and direction

96
00:04:33,031 --> 00:04:36,367
into an ethereal data pigment.

97
00:04:36,583 --> 00:04:38,802
The result was a meditative

98
00:04:38,802 --> 00:04:41,194
yet speculative experience.

99
00:04:41,561 --> 00:04:44,447
This kinetic data sculpture,
titled, "Bosphorus,"

100
00:04:44,447 --> 00:04:48,451
was a similar attempt to question
our capacity to reimagine

101
00:04:48,451 --> 00:04:50,253
natural occurrences.

102
00:04:50,253 --> 00:04:55,969
Using high-frequency radar collections
of the Marmara Sea,

103
00:04:55,969 --> 00:04:58,550
we collected sea surface data

104
00:04:58,550 --> 00:05:02,093
and projected its dynamic movement
with machine intelligence.

105
00:05:02,093 --> 00:05:04,096
We create the sense of immersion

106
00:05:04,096 --> 00:05:06,831
in a calm yet constantly changing

107
00:05:06,831 --> 00:05:08,916
synthetic serial.

108
00:05:09,183 --> 00:05:11,509
Seeing with the brain

109
00:05:11,509 --> 00:05:13,560
is often called imagination,

110
00:05:13,560 --> 00:05:16,064
and, for me, imagining architecture

111
00:05:16,064 --> 00:05:18,348
goes beyond just glass,

112
00:05:18,348 --> 00:05:20,017
metal or concrete,

113
00:05:20,017 --> 00:05:24,494
instead experimenting with
the furthermost possibilities of immersion

114
00:05:24,494 --> 00:05:28,381
and ways of augmenting
our perception in built environments.

115
00:05:28,631 --> 00:05:32,308
Research in artificial intelligence
is growing every day,

116
00:05:32,308 --> 00:05:36,012
leaving us with the feeling
of being plugged into a system

117
00:05:36,012 --> 00:05:38,597
that is bigger and more knowledgeable

118
00:05:38,597 --> 00:05:39,970
than ourselves.

119
00:05:40,199 --> 00:05:43,776
In 2017, we discovered
an open source library

120
00:05:43,776 --> 00:05:46,362
of cultural documents in Istanbul

121
00:05:46,362 --> 00:05:49,998
and began working on "Archive Dreaming,"

122
00:05:49,998 --> 00:05:54,325
one of the first AI-driven
public installations in the world,

123
00:05:54,509 --> 00:06:01,306
an AI exploring approximately
1.7 million documents that span 270 years.

124
00:06:01,888 --> 00:06:04,951
One of our inspirations
during this process

125
00:06:04,951 --> 00:06:08,004
was a short story
called "The Library of Babel"

126
00:06:08,004 --> 00:06:11,656
by the Argentine writer Jorge Luis Borges.

127
00:06:11,656 --> 00:06:17,052
In the story, the author conceives
a universe in the form of a vast library

128
00:06:17,052 --> 00:06:20,254
containing all possible 410-page books

129
00:06:20,254 --> 00:06:22,858
of a certain format and character set.

130
00:06:23,124 --> 00:06:24,815
Through this inspiring image,

131
00:06:24,815 --> 00:06:27,401
we imagine a way to physically explore

132
00:06:27,401 --> 00:06:29,503
the vast archives of knowledge

133
00:06:29,503 --> 00:06:31,788
in the age of machine intelligence.

134
00:06:32,005 --> 00:06:34,040
The resulting work, as you can see,

135
00:06:34,040 --> 00:06:36,615
was a user-driven immersive space.

136
00:06:36,615 --> 00:06:41,838
"Archive Dreaming" profoundly transformed
the experience of a library

137
00:06:41,838 --> 00:06:44,390
in the age of machine intelligence.

138
00:06:44,557 --> 00:06:48,301
"Machine Hallucination"
is an exploration of time and space

139
00:06:48,301 --> 00:06:50,903
experienced through New York City's

140
00:06:50,903 --> 00:06:53,005
public photographic archives.

141
00:06:53,005 --> 00:06:55,481
For this one-of-a-kind immersive project,

142
00:06:55,481 --> 00:06:58,250
we deployed machine-learning algorithms

143
00:06:58,250 --> 00:07:02,321
to find and process over
100 million photographs of the city.

144
00:07:03,188 --> 00:07:06,214
We designed an innovative narrative system

145
00:07:06,214 --> 00:07:12,536
to use artificial intelligence
to predict or to hallucinate new images,

146
00:07:12,536 --> 00:07:16,246
allowing the viewer to step
into a dreamlike fusion

147
00:07:16,246 --> 00:07:18,666
of past and future New York.

148
00:07:20,067 --> 00:07:25,578
As our projects delve deeper into
remembering and transmitting knowledge,

149
00:07:25,578 --> 00:07:28,582
we thought more about how memories

150
00:07:28,582 --> 00:07:30,532
were not static recollections

151
00:07:30,532 --> 00:07:34,369
but ever-changing
interpretations of past events.

152
00:07:34,369 --> 00:07:36,645
We pondered how machines

153
00:07:36,645 --> 00:07:40,282
could simulate unconscious
and subconscious events

154
00:07:40,282 --> 00:07:41,951
such as dreaming,

155
00:07:41,951 --> 00:07:43,300
remembering

156
00:07:43,300 --> 00:07:46,028
and hallucinating.

157
00:07:46,028 --> 00:07:48,296
Thus we created "Melting Memories"

158
00:07:48,296 --> 00:07:51,282
to visualize the moment of remembering.

159
00:07:51,282 --> 00:07:54,869
The inspiration came from a tragic event

160
00:07:54,875 --> 00:07:58,962
when I found out that my uncle
was diagnosed with Alzheimer's.

161
00:07:58,962 --> 00:08:02,282
At that time, all I could think about

162
00:08:02,282 --> 00:08:04,734
was to find a way to celebrate

163
00:08:04,734 --> 00:08:07,429
how and what he remembered

164
00:08:07,429 --> 00:08:09,446
when we are still able to do so.

165
00:08:09,446 --> 00:08:11,911
I began to think of memories

166
00:08:11,911 --> 00:08:13,500
not as disappearing

167
00:08:13,500 --> 00:08:15,101
but as melting

168
00:08:15,101 --> 00:08:16,508
or changing shape.

169
00:08:16,508 --> 00:08:18,377
With the help of machine intelligence,

170
00:08:18,377 --> 00:08:21,580
we worked with the scientists
at the Neuroscape Laboratory

171
00:08:21,864 --> 00:08:23,883
at the University of California,

172
00:08:23,883 --> 00:08:29,444
who showed us how to understand
brain signals as memories are made.

173
00:08:29,444 --> 00:08:34,048
Although my own uncle was losing
the ability to process memories,

174
00:08:34,048 --> 00:08:37,108
the artwork generated by the EEG data

175
00:08:37,392 --> 00:08:41,128
explored the materiality of remembering

176
00:08:41,128 --> 00:08:43,931
and stood as a tribute

177
00:08:43,931 --> 00:08:47,190
to what my uncle had lost.

178
00:08:47,190 --> 00:08:51,611
Almost nothing about contemporary LA

179
00:08:51,611 --> 00:08:54,197
matched my childhood expectation

180
00:08:54,197 --> 00:08:55,538
of the city,

181
00:08:55,538 --> 00:08:58,692
with the exception
of one amazing building:

182
00:08:58,692 --> 00:09:02,311
the Walt Disney Concert Hall,
designed by Frank Gehry,

183
00:09:02,311 --> 00:09:04,380
one of my all-time heroes.

184
00:09:04,380 --> 00:09:07,589
In 2018, I had a call
from the LA Philharmonic

185
00:09:07,589 --> 00:09:09,624
who was looking for an installation

186
00:09:09,624 --> 00:09:13,677
to help mark the celebrated symphony's
hundredth year anniversary.

187
00:09:13,909 --> 00:09:16,988
For this, we decided to ask the question,

188
00:09:16,988 --> 00:09:20,074
can a build learn? Can it dream?

189
00:09:20,825 --> 00:09:22,025
To answer this question,

190
00:09:22,025 --> 00:09:25,402
we decided to collect everything recorded

191
00:09:25,402 --> 00:09:27,923
in the archives of the LA field and WHC,

192
00:09:27,923 --> 00:09:32,811
to be precise, 77 terabytes
of digitally archived memories.

193
00:09:32,811 --> 00:09:35,097
By using machine intelligence,

194
00:09:35,097 --> 00:09:38,433
the entire archive going back 100 years

195
00:09:38,433 --> 00:09:41,319
became projections on the building's skin,

196
00:09:41,319 --> 00:09:45,412
42 projectors to achieve
this futuristic public experience

197
00:09:45,412 --> 00:09:47,664
in the heart of Los Angeles,

198
00:09:47,664 --> 00:09:52,342
getting one step closer
to the LA of "Blade Runner."

199
00:09:52,342 --> 00:09:54,761
If ever a building could dream,

200
00:09:54,761 --> 00:09:56,829
it was in this moment.

201
00:10:00,049 --> 00:10:03,292
Now, I am inviting you to one last journey

202
00:10:03,292 --> 00:10:05,795
into the mind of a machine.

203
00:10:05,795 --> 00:10:09,365
Right now, we are fully immersed
in the data universe

204
00:10:09,365 --> 00:10:12,091
of every single created TED Talk

205
00:10:12,091 --> 00:10:14,092
from the past 30 years.

206
00:10:14,092 --> 00:10:20,565
That means, this data set includes
7,705 talks from the TED stage.

207
00:10:21,433 --> 00:10:23,641
Those talks have been translated

208
00:10:23,641 --> 00:10:25,843
into 7.4 million seconds,

209
00:10:25,843 --> 00:10:30,093
and each second is represented
here in this data universe.

210
00:10:30,093 --> 00:10:32,259
Every image that you are seeing in here

211
00:10:32,259 --> 00:10:35,228
represents unique moments
from those talks.

212
00:10:35,228 --> 00:10:37,221
By using machine intelligence,

213
00:10:37,221 --> 00:10:41,407
we processed a total of 487,000 sentences

214
00:10:41,407 --> 00:10:45,819
into 330 unique clusters of topics
like nature, global emissions,

215
00:10:45,819 --> 00:10:47,620
extinction, race issues,

216
00:10:47,620 --> 00:10:49,105
computation,

217
00:10:49,105 --> 00:10:52,758
trust, emotions, water and refugees.

218
00:10:52,758 --> 00:10:55,630
These clusters are then
connected to each other

219
00:10:55,630 --> 00:10:56,784
by an algorithm,

220
00:10:56,784 --> 00:11:00,950
and generated 113 million line segments,

221
00:11:00,950 --> 00:11:03,910
which revealed new
conceptual relationships.

222
00:11:03,910 --> 00:11:06,064
Wouldn't it be amazing

223
00:11:06,064 --> 00:11:07,381
to be able to remember

224
00:11:07,381 --> 00:11:10,450
all the questions that have ever
been asked on the stage?

225
00:11:11,758 --> 00:11:13,126
Here I am,

226
00:11:13,126 --> 00:11:16,011
inside the mind
of countless great thinkers,

227
00:11:16,011 --> 00:11:17,629
as well as a machine

228
00:11:17,629 --> 00:11:20,032
interacting with various feelings

229
00:11:20,032 --> 00:11:22,241
attributed to learning,

230
00:11:22,241 --> 00:11:23,325
remembering,

231
00:11:23,325 --> 00:11:24,460
questioning

232
00:11:24,460 --> 00:11:27,213
and imagining all at the same time,

233
00:11:27,529 --> 00:11:30,882
expanding the power of the mind.

234
00:11:31,198 --> 00:11:33,691
For me, being right here

235
00:11:33,691 --> 00:11:38,113
is indeed what it means
to be an AI in the 21st century.

236
00:11:38,113 --> 00:11:40,515
It is in our hands, humans,

237
00:11:40,515 --> 00:11:44,141
to train this mind to learn and remember

238
00:11:44,141 --> 00:11:46,994
what we can only dream of.

239
00:11:46,994 --> 00:11:48,362
Thank you.