1
00:00:00,784 --> 00:00:03,659
Hi, I'm Refik. I'm a media artist.

2
00:00:03,683 --> 00:00:05,658
I use data as a pigment

3
00:00:05,682 --> 00:00:07,572
and paint with a thinking brush

4
00:00:07,596 --> 00:00:10,455
that is assisted
by artificial intelligence.

5
00:00:11,128 --> 00:00:13,876
Using architectural spaces as canvases,

6
00:00:13,900 --> 00:00:15,694
I collaborate with machines

7
00:00:15,718 --> 00:00:18,661
to make buildings dream and hallucinate.

8
00:00:18,685 --> 00:00:21,400
You may be wondering,
what does all this mean?

9
00:00:21,424 --> 00:00:25,089
So let me please take you
into my work and my world.

10
00:00:25,851 --> 00:00:29,590
I witnessed the power of imagination
when I was eight years old,

11
00:00:29,614 --> 00:00:31,956
as a child growing up in Istanbul.

12
00:00:31,980 --> 00:00:35,371
One day, my mom brought home
a videocassette

13
00:00:35,395 --> 00:00:38,078
of the science-fiction movie
"Blade Runner."

14
00:00:38,102 --> 00:00:41,077
I clearly remember being mesmerized

15
00:00:41,101 --> 00:00:46,010
by the stunning architectural vision
of the future of Los Angeles,

16
00:00:46,034 --> 00:00:48,586
a place that I had never seen before.

17
00:00:48,610 --> 00:00:54,007
That vision became
a kind of a staple of my daydreams.

18
00:00:54,031 --> 00:00:56,793
When I arrived in LA in 2012

19
00:00:56,817 --> 00:00:59,237
for a graduate program
in Design Media Arts,

20
00:00:59,261 --> 00:01:01,651
I rented a car and drove downtown

21
00:01:01,675 --> 00:01:04,721
to see that wonderful world
of the near future.

22
00:01:05,482 --> 00:01:07,128
I remember a specific line

23
00:01:07,152 --> 00:01:10,221
that kept playing
over and over in my head:

24
00:01:10,245 --> 00:01:12,340
the scene when the android Rachael

25
00:01:12,364 --> 00:01:16,079
realizes that her memories
are actually not hers,

26
00:01:16,103 --> 00:01:20,754
and when Deckard tells her
they are someone else's memories.

27
00:01:20,778 --> 00:01:22,011
Since that moment,

28
00:01:22,035 --> 00:01:24,915
one of my inspirations
has been this question.

29
00:01:25,561 --> 00:01:29,900
What can a machine do
with someone else's memories?

30
00:01:29,924 --> 00:01:32,627
Or, to say that in another way,

31
00:01:32,651 --> 00:01:36,402
what does it mean to be an AI
in the 21st century?

32
00:01:37,441 --> 00:01:39,538
Any android or AI machine

33
00:01:39,562 --> 00:01:42,679
is only intelligent
as long as we collaborate with it.

34
00:01:43,327 --> 00:01:44,792
It can construct things

35
00:01:44,816 --> 00:01:48,072
that human intelligence intends to produce

36
00:01:48,096 --> 00:01:50,238
but does not have the capacity to do so.

37
00:01:51,373 --> 00:01:55,178
Think about your activities
and social networks, for example.

38
00:01:55,208 --> 00:01:58,756
They get smarter
the more you interact with them.

39
00:01:58,780 --> 00:02:03,249
If machines can learn or process memories,

40
00:02:03,273 --> 00:02:05,134
can they also dream?

41
00:02:05,158 --> 00:02:06,644
Hallucinate?

42
00:02:06,668 --> 00:02:09,082
Involuntarily remember,

43
00:02:09,106 --> 00:02:13,059
or make connections
between multiple people's dreams?

44
00:02:13,083 --> 00:02:19,106
Does being an AI in the 21st century
simply mean not forgetting anything?

45
00:02:20,365 --> 00:02:21,548
And, if so,

46
00:02:21,572 --> 00:02:25,826
isn't it the most revolutionary thing
that we have experienced

47
00:02:25,850 --> 00:02:30,762
in our centuries-long effort
to capture history across media?

48
00:02:31,406 --> 00:02:32,740
In other words,

49
00:02:32,764 --> 00:02:36,217
how far have we come
since Ridley Scott's "Blade Runner"?

50
00:02:36,678 --> 00:02:40,165
So I established my studio in 2014

51
00:02:40,189 --> 00:02:42,200
and invited architects,

52
00:02:42,224 --> 00:02:44,596
computer and data scientists,
neuroscientists,

53
00:02:44,620 --> 00:02:47,467
musicians and even storytellers

54
00:02:47,491 --> 00:02:50,232
to join me in realizing my dreams.

55
00:02:51,001 --> 00:02:53,835
Can data become a pigment?

56
00:02:53,859 --> 00:02:56,446
This was the very first question we asked

57
00:02:56,470 --> 00:03:01,017
when starting our journey
to embed media arts into architecture,

58
00:03:01,041 --> 00:03:03,699
to collide virtual and physical worlds.

59
00:03:04,555 --> 00:03:09,294
So we began to imagine
what I would call the poetics of data.

60
00:03:10,124 --> 00:03:12,841
One of our first projects,
"Virtual Depictions,"

61
00:03:12,865 --> 00:03:14,964
was a public data sculpture piece

62
00:03:14,988 --> 00:03:17,688
commissioned by the city of San Francisco.

63
00:03:17,712 --> 00:03:19,918
The work invites the audience

64
00:03:19,942 --> 00:03:23,157
to be part of a spectacular
aesthetic experience

65
00:03:23,181 --> 00:03:24,820
in a living urban space

66
00:03:24,844 --> 00:03:30,415
by depicting a fluid network
of connections of the city itself.

67
00:03:30,439 --> 00:03:33,162
It also stands as a reminder

68
00:03:33,186 --> 00:03:36,546
of how invisible data
from our everyday lives,

69
00:03:36,570 --> 00:03:39,736
like the Twitter feeds
that are represented here,

70
00:03:39,760 --> 00:03:41,888
can be made visible

71
00:03:41,912 --> 00:03:47,609
and transformed into sensory knowledge
that can be experienced collectively.

72
00:03:48,619 --> 00:03:53,519
In fact, data can only become knowledge
when it's experienced,

73
00:03:53,543 --> 00:03:57,519
and what is knowledge and experience
can take many forms.

74
00:03:57,543 --> 00:03:59,303
When exploring such connections

75
00:03:59,327 --> 00:04:03,833
through the vast potential
of machine intelligence,

76
00:04:03,857 --> 00:04:09,072
we also pondered the connection
between human senses

77
00:04:09,096 --> 00:04:12,747
and the machines' capacity
for simulating nature.

78
00:04:12,771 --> 00:04:17,771
These inquiries began
while working on wind-data paintings.

79
00:04:17,795 --> 00:04:20,755
They took the shape of visualized poems

80
00:04:20,779 --> 00:04:25,593
based on hidden data sets
that we collected from wind sensors.

81
00:04:25,617 --> 00:04:28,430
We then used generative algorithms

82
00:04:28,454 --> 00:04:32,645
to transform wind speed,
gust and direction

83
00:04:32,669 --> 00:04:35,303
into an ethereal data pigment.

84
00:04:36,387 --> 00:04:40,418
The result was a meditative
yet speculative experience.

85
00:04:41,349 --> 00:04:44,212
This kinetic data sculpture,
titled "Bosphorus,"

86
00:04:44,236 --> 00:04:48,553
was a similar attempt to question
our capacity to reimagine

87
00:04:48,577 --> 00:04:49,982
natural occurrences.

88
00:04:51,299 --> 00:04:55,736
Using high-frequency radar collections
of the Marmara Sea,

89
00:04:55,760 --> 00:04:57,990
we collected sea-surface data

90
00:04:58,014 --> 00:05:01,195
and projected its dynamic movement
with machine intelligence.

91
00:05:01,874 --> 00:05:03,972
We create a sense of immersion

92
00:05:03,996 --> 00:05:08,175
in a calm yet constantly changing
synthetic sea view.

93
00:05:09,524 --> 00:05:13,612
Seeing with the brain
is often called imagination,

94
00:05:13,636 --> 00:05:15,940
and, for me, imagining architecture

95
00:05:15,964 --> 00:05:19,845
goes beyond just glass, metal or concrete,

96
00:05:19,869 --> 00:05:24,370
instead experimenting with
the furthermost possibilities of immersion

97
00:05:24,394 --> 00:05:28,507
and ways of augmenting
our perception in built environments.

98
00:05:28,531 --> 00:05:32,184
Research in artificial intelligence
is growing every day,

99
00:05:32,208 --> 00:05:36,086
leaving us with the feeling
of being plugged into a system

100
00:05:36,110 --> 00:05:38,404
that is bigger and more knowledgeable

101
00:05:38,428 --> 00:05:39,880
than ourselves.

102
00:05:39,904 --> 00:05:43,508
In 2017, we discovered
an open-source library

103
00:05:43,532 --> 00:05:46,174
of cultural documents in Istanbul

104
00:05:46,198 --> 00:05:49,658
and began working on "Archive Dreaming,"

105
00:05:49,682 --> 00:05:53,976
one of the first AI-driven
public installations in the world,

106
00:05:54,000 --> 00:06:00,975
an AI exploring approximately
1.7 million documents that span 270 years.

107
00:06:01,788 --> 00:06:04,630
One of our inspirations
during this process

108
00:06:04,654 --> 00:06:08,081
was a short story
called "The Library of Babel"

109
00:06:08,105 --> 00:06:11,312
by the Argentine writer Jorge Luis Borges.

110
00:06:11,336 --> 00:06:17,065
In the story, the author conceives
a universe in the form of a vast library

111
00:06:17,089 --> 00:06:23,000
containing all possible 410-page books
of a certain format and character set.

112
00:06:23,024 --> 00:06:24,493
Through this inspiring image,

113
00:06:24,517 --> 00:06:29,247
we imagine a way to physically explore
the vast archives of knowledge

114
00:06:29,271 --> 00:06:31,762
in the age of machine intelligence.

115
00:06:31,786 --> 00:06:33,844
The resulting work, as you can see,

116
00:06:33,868 --> 00:06:36,491
was a user-driven immersive space.

117
00:06:36,515 --> 00:06:41,390
"Archive Dreaming" profoundly transformed
the experience of a library

118
00:06:41,414 --> 00:06:44,163
in the age of machine intelligence.

119
00:06:44,187 --> 00:06:48,255
"Machine Hallucination"
is an exploration of time and space

120
00:06:48,279 --> 00:06:52,732
experienced through New York City's
public photographic archives.

121
00:06:52,756 --> 00:06:55,382
For this one-of-a-kind immersive project,

122
00:06:55,406 --> 00:06:58,070
we deployed machine-learning algorithms

123
00:06:58,094 --> 00:07:02,256
to find and process over
100 million photographs of the city.

124
00:07:03,008 --> 00:07:06,090
We designed an innovative narrative system

125
00:07:06,114 --> 00:07:12,317
to use artificial intelligence
to predict or to hallucinate new images,

126
00:07:12,341 --> 00:07:16,264
allowing the viewer
to step into a dreamlike fusion

127
00:07:16,288 --> 00:07:18,412
of past and future New York.

128
00:07:19,832 --> 00:07:21,651
As our projects delve deeper

129
00:07:21,675 --> 00:07:25,351
into remembering
and transmitting knowledge,

130
00:07:25,375 --> 00:07:30,333
we thought more about how memories
were not static recollections

131
00:07:30,357 --> 00:07:33,718
but ever-changing
interpretations of past events.

132
00:07:34,269 --> 00:07:36,350
We pondered how machines

133
00:07:36,374 --> 00:07:40,404
could simulate unconscious
and subconscious events,

134
00:07:40,428 --> 00:07:44,340
such as dreaming,
remembering and hallucinating.

135
00:07:45,356 --> 00:07:48,260
Thus, we created "Melting Memories"

136
00:07:48,284 --> 00:07:50,705
to visualize the moment of remembering.

137
00:07:51,824 --> 00:07:54,491
The inspiration came from a tragic event,

138
00:07:54,515 --> 00:07:58,026
when I found out that my uncle
was diagnosed with Alzheimer's.

139
00:07:59,602 --> 00:08:02,023
At that time, all I could think about

140
00:08:02,047 --> 00:08:07,110
was to find a way to celebrate
how and what we remember

141
00:08:07,134 --> 00:08:09,128
when we are still able to do so.

142
00:08:09,152 --> 00:08:13,232
I began to think of memories
not as disappearing

143
00:08:13,256 --> 00:08:16,168
but as melting or changing shape.

144
00:08:16,192 --> 00:08:18,253
With the help of machine intelligence,

145
00:08:18,277 --> 00:08:21,740
we worked with the scientists
at the Neuroscape Laboratory

146
00:08:21,764 --> 00:08:23,603
at the University of California,

147
00:08:23,627 --> 00:08:29,184
who showed us how to understand
brain signals as memories are made.

148
00:08:29,208 --> 00:08:34,311
Although my own uncle was losing
the ability to process memories,

149
00:08:34,335 --> 00:08:37,858
the artwork generated by EEG data

150
00:08:37,882 --> 00:08:41,184
explored the materiality of remembering

151
00:08:41,208 --> 00:08:45,588
and stood as a tribute
to what my uncle had lost.

152
00:08:48,533 --> 00:08:51,487
Almost nothing about contemporary LA

153
00:08:51,511 --> 00:08:55,272
matched my childhood
expectation of the city,

154
00:08:55,296 --> 00:08:58,386
with the exception
of one amazing building:

155
00:08:58,410 --> 00:09:01,799
the Walt Disney Concert Hall,
designed by Frank Gehry,

156
00:09:01,823 --> 00:09:03,554
one of my all-time heroes.

157
00:09:04,208 --> 00:09:07,565
In 2018, I had a call
from the LA Philharmonic

158
00:09:07,589 --> 00:09:09,500
who was looking for an installation

159
00:09:09,524 --> 00:09:13,614
to help mark the celebrated symphony's
hundred-year anniversary.

160
00:09:13,638 --> 00:09:17,010
For this, we decided to ask the question,

161
00:09:17,034 --> 00:09:19,974
"Can a building learn? Can it dream?"

162
00:09:20,614 --> 00:09:21,791
To answer this question,

163
00:09:21,815 --> 00:09:27,646
we decided to collect everything recorded
in the archives of the LA Phil and WDCH.

164
00:09:27,670 --> 00:09:32,535
To be precise, 77 terabytes
of digitally archived memories.

165
00:09:32,559 --> 00:09:34,726
By using machine intelligence,

166
00:09:34,750 --> 00:09:38,166
the entire archive, going back 100 years,

167
00:09:38,190 --> 00:09:41,195
became projections on the building's skin,

168
00:09:41,219 --> 00:09:45,163
42 projectors to achieve
this futuristic public experience

169
00:09:45,187 --> 00:09:47,362
in the heart of Los Angeles,

170
00:09:47,386 --> 00:09:51,369
getting one step closer
to the LA of "Blade Runner."

171
00:09:52,146 --> 00:09:54,660
If ever a building could dream,

172
00:09:54,684 --> 00:09:56,361
it was in this moment.

173
00:09:59,703 --> 00:10:04,449
Now, I am inviting you to one last journey
into the mind of a machine.

174
00:10:05,877 --> 00:10:09,241
Right now, we are fully immersed
in the data universe

175
00:10:09,265 --> 00:10:13,813
of every single curated TED Talk
from the past 30 years.

176
00:10:13,837 --> 00:10:20,439
That means this data set includes
7,705 talks from the TED stage.

177
00:10:21,094 --> 00:10:25,635
Those talks have been translated
into 7.4 million seconds,

178
00:10:25,659 --> 00:10:29,754
and each second is represented
here in this data universe.

179
00:10:29,778 --> 00:10:31,849
Every image that you are seeing in here

180
00:10:31,873 --> 00:10:34,945
represents unique moments
from those talks.

181
00:10:34,969 --> 00:10:36,842
By using machine intelligence,

182
00:10:36,866 --> 00:10:41,167
we processed a total of 487,000 sentences

183
00:10:41,191 --> 00:10:45,580
into 330 unique clusters of topics
like nature, global emissions,

184
00:10:45,604 --> 00:10:48,890
extinction, race issues, computation,

185
00:10:48,914 --> 00:10:52,454
trust, emotions, water and refugees.

186
00:10:52,478 --> 00:10:55,327
These clusters are then
connected to each other

187
00:10:55,351 --> 00:10:56,613
by an algorithm,

188
00:10:56,637 --> 00:11:00,479
[that] generated 113 million
line segments,

189
00:11:00,503 --> 00:11:03,606
which reveal new conceptual relationships.

190
00:11:03,630 --> 00:11:06,971
Wouldn't it be amazing
to be able to remember

191
00:11:06,995 --> 00:11:10,064
all the questions that have ever
been asked on the stage?

192
00:11:11,507 --> 00:11:12,946
Here I am,

193
00:11:12,970 --> 00:11:15,832
inside the mind
of countless great thinkers,

194
00:11:15,856 --> 00:11:19,579
as well as a machine,
interacting with various feelings

195
00:11:19,603 --> 00:11:21,968
attributed to learning,

196
00:11:21,992 --> 00:11:24,228
remembering, questioning

197
00:11:24,252 --> 00:11:27,619
and imagining all at the same time,

198
00:11:27,643 --> 00:11:29,918
expanding the power of the mind.

199
00:11:31,034 --> 00:11:33,431
For me, being right here

200
00:11:33,455 --> 00:11:37,407
is indeed what it means
to be an AI in the 21st century.

201
00:11:38,184 --> 00:11:40,477
It is in our hands, humans,

202
00:11:40,501 --> 00:11:44,017
to train this mind to learn and remember

203
00:11:44,041 --> 00:11:46,134
what we can only dream of.

204
00:11:47,258 --> 00:11:48,408
Thank you.