1
00:00:00,120 --> 00:00:04,220
Dear Florian,
my apologies for not joining today.

2
00:00:04,250 --> 00:00:09,220
While you're listening to this, I am
doing another presentation across town.

3
00:00:09,250 --> 00:00:15,380
No doubt in the not so diststant future,
I'll be able to be in two places at once.

4
00:00:15,410 --> 00:00:17,620
What was sneered at as deep fake

5
00:00:17,650 --> 00:00:21,940
with the evolution of AI will simply be
replicated iterations of the self

6
00:00:21,970 --> 00:00:25,540
that keep us performing everything,
everywhere, all at once.

7
00:00:25,570 --> 00:00:28,260
It's a neoliberal, if not colonial,

8
00:00:28,290 --> 00:00:33,420
dream come true of 24/7 extraction
across all longitudes and latitude.

9
00:00:33,450 --> 00:00:35,100
The sun will never set on this

10
00:00:35,130 --> 00:00:40,220
computational empire, where data is
mined and CPUs endlessly render.

11
00:00:40,250 --> 00:00:41,340
FOMO.

12
00:00:41,370 --> 00:00:45,900
That abbreviation for the fear of missing
out will be neatly filed away in what

13
00:00:45,930 --> 00:00:51,260
the science fiction writer Bruce Sterling
named a dead media project

14
00:00:51,290 --> 00:00:54,620
with the ubiquitous presence
of our proliferated selves.

15
00:00:54,650 --> 00:00:57,300
FOMO, along with dialup modems.

16
00:00:57,330 --> 00:00:58,980
Welcome to my home page.

17
00:00:59,010 --> 00:01:01,860
Floppy disks and MySpace will be nothing

18
00:01:01,890 --> 00:01:06,620
but nostalgia for technologists
and fodder for media archaeologists.

19
00:01:06,650 --> 00:01:10,700
But for the moment,
let's remain in this present while knowing

20
00:01:10,730 --> 00:01:14,260
we're always in the archive
with history nipping at our heels.

21
00:01:14,290 --> 00:01:16,100
You asked whether I could give a few

22
00:01:16,130 --> 00:01:20,660
reflections on AI, and to do
so I begin with a disclaimer.

23
00:01:20,690 --> 00:01:22,780
I am not an expert in the field.

24
00:01:22,810 --> 00:01:28,340
My perspective is that of a mother,
teacher, learner, artist, gardener, cook,

25
00:01:28,370 --> 00:01:31,780
lover, Netflix, watcher,
and Internet user.

26
00:01:31,810 --> 00:01:36,040
A computer engineer or economist would
undoubtedly lend insights regarding

27
00:01:36,070 --> 00:01:41,380
the infrastructural, programmatic and
financial underpinnings integral to AI.

28
00:01:41,410 --> 00:01:45,100
As we both know,
matter matters form informs,

29
00:01:45,130 --> 00:01:49,210
and you've got to follow the money
to understand where power resides.

30
00:01:49,240 --> 00:01:53,380
Nonetheless, AI is embedded
in my daily routine.

31
00:01:53,410 --> 00:01:55,820
It permeates my Google searches,

32
00:01:55,850 --> 00:02:01,020
suggests autogenerated responses
in Outlook, gives Amazon recommendations,

33
00:02:01,050 --> 00:02:05,100
provides grammatical corrections
in grammarly, and tells me I might like

34
00:02:05,130 --> 00:02:10,220
season five of The Handmaid's Tale,
even though I've never seen season one.

35
00:02:10,250 --> 00:02:11,940
Stalking my activities,

36
00:02:11,970 --> 00:02:17,140
the machines munch and crunch, ingesting
and combining my data with others.

37
00:02:17,170 --> 00:02:23,300
It is a mirror, a projection,
a mirage, a prompt, and a directive.

38
00:02:23,330 --> 00:02:25,780
A profile can be viewed as a composite.

39
00:02:25,810 --> 00:02:28,700
Multitudes, glued together like a collage.

40
00:02:28,730 --> 00:02:33,980
But rather than a composite, what if these
processes were more akin to composting?

41
00:02:34,010 --> 00:02:39,170
As I write this message, I'm sitting in
my garden, looking at my vegetable beds.

42
00:02:39,200 --> 00:02:40,700
Once my courgettes,

43
00:02:40,730 --> 00:02:45,540
pumpkins and beans have gone through their
cycle, the leaves and stems are put

44
00:02:45,570 --> 00:02:50,380
into the compost and eventually provide
nutrients for the following year.

45
00:02:50,410 --> 00:02:54,420
Every gardener knows the significance
of soil and compost.

46
00:02:54,450 --> 00:02:58,980
The golden rule is it's only as good as
what goes into it,

47
00:02:59,010 --> 00:03:02,580
and as the pile is sifted
and turned from one box to another.

48
00:03:02,610 --> 00:03:08,260
A gardener also knows compost is not just
fertilizer, but the future in the making.

49
00:03:08,290 --> 00:03:12,220
Like those who plant trees,
they understand themselves within a longer

50
00:03:12,250 --> 00:03:16,060
timeline, a continuum that is
greater than their own.

51
00:03:16,090 --> 00:03:18,640
Sadly, I suspect datasets and their

52
00:03:18,670 --> 00:03:22,060
algorithmic perimeters don't
get the same attention.

53
00:03:22,090 --> 00:03:26,260
Instead, data is just another
resource to be exploited.

54
00:03:26,290 --> 00:03:29,980
Like fossil fuel it is
something to be mined.

55
00:03:30,010 --> 00:03:31,880
Unlike the permaculture paradigm

56
00:03:31,910 --> 00:03:36,980
of renewal and regeneration, this
worldview is one of total consumption.

57
00:03:37,010 --> 00:03:42,180
In this model, the motto is, It's only
as good as what we can get out of it.

58
00:03:42,210 --> 00:03:44,840
And as the Indian writer Amitav Ghosh

59
00:03:44,870 --> 00:03:49,780
warns in his book "The Nut makes curse,
parables for a planet in crisis",

60
00:03:49,810 --> 00:03:54,980
the view of the world as resource has
deep and destructive colonial roots.

61
00:03:55,010 --> 00:03:59,380
Recounting the history of the VOCs
monopolization of the Nutmeg trade

62
00:03:59,410 --> 00:04:04,260
and the subsequent slaughter
and eradication of the Bandanese in 1621,

63
00:04:04,290 --> 00:04:09,580
he exposes how these legacies are directly
linked to our current climate crisis.

64
00:04:09,610 --> 00:04:14,940
Bersch contrasts this colonial domination
and catastrophic violence with indigenous

65
00:04:14,970 --> 00:04:20,420
knowledge, which values the human and more
than human, the animate and socalled

66
00:04:20,450 --> 00:04:25,660
inanimate as equally vibrant and,
most importantly, interdependent.

67
00:04:25,690 --> 00:04:27,860
Lately, I've been wondering if maybe

68
00:04:27,890 --> 00:04:32,220
the word artificial blinds us
to the reality that AI is of this Earth.

69
00:04:32,250 --> 00:04:37,940
Despite the deceptive moniker,
cloud computing is far from ethereal.

70
00:04:37,970 --> 00:04:43,540
It is matiere, that is,
it is made of material substance.

71
00:04:43,570 --> 00:04:45,560
System boards are forged from metals

72
00:04:45,590 --> 00:04:49,020
and minerals, processors need enormous
amounts of energy,

73
00:04:49,040 --> 00:04:53,540
and ever expansive server farms
require continuous water cooling.

74
00:04:53,570 --> 00:04:55,380
To sustain these systems,

75
00:04:55,410 --> 00:04:59,860
extraction happens on multiple
and interconnected levels.

76
00:04:59,890 --> 00:05:00,980
Some time ago,

77
00:05:01,010 --> 00:05:06,780
I read that Google considers its water
usage a proprietary trade secret and bars

78
00:05:06,800 --> 00:05:10,700
even public officials from disclosing
the company's consumption.

79
00:05:10,720 --> 00:05:16,700
But from legal cases in 2019 alone,
it was surmised that over 2.3 billion

80
00:05:16,720 --> 00:05:20,140
gallons of water were being
used across three states.

81
00:05:20,160 --> 00:05:22,420
One of those states, Texas,

82
00:05:22,450 --> 00:05:27,100
where I'm from, has suffered
unprecedented drought in the past years.

83
00:05:27,130 --> 00:05:30,860
I can't help but ask myself,
how can something essential to our

84
00:05:30,890 --> 00:05:34,660
existence beat Google's
proprietary trade secret?

85
00:05:34,690 --> 00:05:38,900
Clearly, this munching and crunching
has an environmental impact.

86
00:05:38,920 --> 00:05:41,620
At this point, you might think I'm going

87
00:05:41,650 --> 00:05:46,180
to make a luddite argument for turning
back the clock, closing Pandora's box,

88
00:05:46,210 --> 00:05:49,100
or putting a spanner
in the wheels of acceleration.

89
00:05:49,130 --> 00:05:50,540
But I'm not.

90
00:05:50,570 --> 00:05:53,420
I'm suspicious of such positions.

91
00:05:53,450 --> 00:05:56,020
Technology never uninvents itself.

92
00:05:56,040 --> 00:05:58,420
It only becomes obsolete.

93
00:05:58,450 --> 00:06:02,460
With this in mind, I return
to my humble compost pile.

94
00:06:02,480 --> 00:06:07,580
It's lessons and the golden rule of "it's
only as good as what goes into it",

95
00:06:07,600 --> 00:06:11,580
as opposed to "It's only as good
as what we can get out of it".

96
00:06:11,600 --> 00:06:17,020
Or, as Donna Haraway writes,
I compost my soul in this hot pile.

97
00:06:17,040 --> 00:06:18,940
The worms are not human,

98
00:06:18,970 --> 00:06:24,620
their undulating bodies ingest and reach,
and their feces fertilize worlds.

99
00:06:24,650 --> 00:06:27,700
AI is a part of the pile too.

100
00:06:27,730 --> 00:06:31,900
Many have said this before,
and much better than I can.

101
00:06:31,920 --> 00:06:36,220
We need to think outside the longstanding
beliefs of our own exceptionalism,

102
00:06:36,250 --> 00:06:41,060
acknowledge complexity
and interdependency, and consider the long

103
00:06:41,090 --> 00:06:46,140
view this planet is a wondrous
and precious place to be,

104
00:06:46,170 --> 00:06:51,100
and in our compost, we must think about
what futures we bring to fruition.

105
00:06:51,130 --> 00:06:53,020
In her book Dear Science,

106
00:06:53,040 --> 00:06:57,860
catherine Mcitrich writes about predictive
algorithms, saying they are anticipatory

107
00:06:57,890 --> 00:07:00,380
computations that tell
us what we already know.

108
00:07:00,410 --> 00:07:01,940
But in the future.

109
00:07:01,970 --> 00:07:04,580
If we want different or better or more

110
00:07:04,600 --> 00:07:08,220
just futures and worlds,
it is important to notice what kind

111
00:07:08,250 --> 00:07:11,740
of knowledge networks are
already predicting our futures.

112
00:07:11,770 --> 00:07:13,900
McKitrick is right.

113
00:07:13,920 --> 00:07:15,620
As I mentioned earlier,

114
00:07:15,650 --> 00:07:19,940
we're always in the archive with history
nipping at our heels,

115
00:07:19,970 --> 00:07:23,420
immersed and implicated without
any authoritative view.

116
00:07:23,450 --> 00:07:25,740
The question is how to proceed consciously

117
00:07:25,770 --> 00:07:29,460
and critically with care,
with a sense of tending to and with an eye

118
00:07:29,480 --> 00:07:33,300
on what might be at stake
rather than training AI.

119
00:07:33,330 --> 00:07:35,860
What if education was emphasized?

120
00:07:35,890 --> 00:07:40,580
It might sound odd in this context,
but if we don't, I believe we'll never

121
00:07:40,600 --> 00:07:45,020
witness any intelligence,
and only the artificial will remain.

122
00:07:45,040 --> 00:07:47,380
Please stop.

123
00:07:47,410 --> 00:07:51,620
Whatever you do, don't put
that garbage in the compost.

124
00:07:51,650 --> 00:07:55,220
If we think about AI within the framework
of education,

125
00:07:55,250 --> 00:08:00,020
critical questions can be raised about its
curriculum, pedagogical approaches

126
00:08:00,040 --> 00:08:02,980
that might be required,
and the disciplinary frameworks that are

127
00:08:03,010 --> 00:08:06,820
needed to enrich knowledge beyond
what is solely instrumental.

128
00:08:06,850 --> 00:08:09,020
We can start to question who its teachers

129
00:08:09,040 --> 00:08:13,700
are, what biases are being concretized,
what the assessment criteria of its

130
00:08:13,730 --> 00:08:17,260
successes and failures are,
and according to whom.

131
00:08:17,290 --> 00:08:22,180
Playing around for this presentation,
I spent several hours testing Dall-E 2

132
00:08:22,210 --> 00:08:26,320
which, according to their marketing
description, is a new AI system that can

133
00:08:26,350 --> 00:08:31,220
create realistic images and art
from a description in natural language.

134
00:08:31,250 --> 00:08:33,880
I entered the simple sentence "she sits

135
00:08:33,910 --> 00:08:37,820
in her garden writing a letter
while wearing a straw hat".

136
00:08:37,850 --> 00:08:40,420
All the results, and there were over 100

137
00:08:40,440 --> 00:08:44,660
of them, were images of white
women sitting in gardens.

138
00:08:44,690 --> 00:08:46,820
I suppose they've never read the antigen

139
00:08:46,850 --> 00:08:50,860
american writer and gardener
Jamaica Kinkade.

140
00:08:50,890 --> 00:08:52,380
I thought to myself,

141
00:08:52,410 --> 00:08:56,500
it's the manifestation of such
an impoverished imagination.

142
00:08:56,530 --> 00:09:00,660
But is this the fault of the machines,
datasets and algorithms?

143
00:09:00,690 --> 00:09:02,820
But of those who educate them?

144
00:09:02,850 --> 00:09:07,500
I often think about the fact
that Alan Turing's 1950 seminal essay,

145
00:09:07,530 --> 00:09:12,820
computing Machinery and Intelligence was
published in "Mind a quarterly review

146
00:09:12,850 --> 00:09:16,420
of Philosophy", and not
a computer science journal.

147
00:09:16,440 --> 00:09:18,500
And as we abandon AI to corporate

148
00:09:18,530 --> 00:09:22,180
interests, I think there is a profound
message in his choice,

149
00:09:22,200 --> 00:09:26,060
which is well worth mulling over as we
churn together in the compost of our

150
00:09:26,080 --> 00:09:30,340
present and create toxins or
nutrients for our future.

151
00:09:30,370 --> 00:09:33,100
There are choices to be made of desolation

152
00:09:33,130 --> 00:09:37,220
and neglect, or of love,
stewardship and care.

153
00:09:37,250 --> 00:09:38,840
Much affection, Renee.