MEMORY,
DRAW
Lev Manovich
solo exhibition | Seoul, South Korea
12/20/2023 - 02/16/2024
Catalog text and design: Lev Manovich
[Link]
ABOUT THE EXHIBITION
The exhibition presents a selection of my digital images from
2023. The title of the exhibition refers to the use of
generative AI as a "memory machine." AI models extract
information from existing cultural material during the
training process, constructing a new historical archive. And
when we probe inside this archive, asking AI tools to
generate scenes from a specific historical period or place, or
to create something unique, the results are frequently
idealized, generic imagery, or bland. For me, using
generative AI is a constant battle against this fundamental
limitation of this medium. The introductory texts below
describe some of the techniques I employ to push AI to
generate more subtle, delicate, and ambitious images that I
prefer.
The title of the exhibition also refers to a number of my
works on paper from the early 1980s, which are displayed as
digital prints in this exhibition. These drawings were created
following my emigration from the Soviet Union in 1981 (see
the "Closed Word" section of the catalog).
Finally, the exhibition title alludes to "Memory, Speak," the
title of a 1951 memoir by Vladimir Nabokov, a famous
Russian emigrant writer.
IN THE GARDEN
Digital images created with generative AI and edited in
Lightroom, 2023.
This series is inspired by formal French gardens as well as
17th-18th century architecture and ornament. I also wanted
AI to simulate the look of old engravings and etchings. The
reference to this medium in my prompts influenced the
subjects depicted in the generated images. Lines represent
everything in graphic arts, and the world itself is made up of
them in my images: curvy ones for plants, straight ones for
poles, and diagonal ones for wires stretched between them.
The series also makes use of the specific artifacts of the AI
"zoom out" command: symmetries and repetitions of
compositional elements. This seemed appropriate given the
topic of formal gardens. In some images, the initial world has
been enclosed by elaborate frames that are nested inside
other frames.
A PLAY FOR A
WINTER CITY
Digital images created with generative AI and edited in
Lightroom, 2022.
The descriptions of the images I gave to AI to create this series
specify “architecture model” or “theatre set design.” This is
one of the techniques I used to push AI away from its default
photo realism towards a different aesthetics where
imperfections looks motivated. In an animated films we may
see simplified characters and minimal features of and space -
and this is what I wanted to get.
The imaginary city in this series is made from simulations of
memories. The city is inhabited with atmosphere, moods, the
particular light and particular cold of the winters in my
childhood. They only now exist in my imagination. Although all
details shown by AI such as people or buildings are not
precise, the atmosphere and the feeling of a winter in the
1970s - exaggerated by the visual references to animated
films from Eastern Europe made at that time - feels right to
me.
LIBRARY OF
UNWRITTEN
MANUSCRIPTS
Digital images created with generative AI and edited in
Lightroom, 2022.
Usually libraries collect already published books. But let’s ask
a question, using the history of literature in 20th century
USSR as example. Do the works of fiction and that were
published in USSR include best novels, best poems, best
most innovative ideas? Often the answer is no. Many of the
most talented and important writes in Russia could not
publish their best works during this period - and sometimes
no works at all. A few had their work published in other
countries.
But many works remained only in manuscript forms. Or they
were not even written down. Instead friends memorized
them and wrote down many years later. But the very best
literary works were not even written down. They remained in
writers’ heads. As plans. And perhaps even as detailed
outlines or finished parts - which the authors never
committed to papers in fear, or simply knowing that they will
be never published in USSR.
This history is the background for Library of Unwritten
Manuscripts series. It is an imaginary library that only houses
writers’ manuscripts of literary works that were either 1) never
published, or 2) never completed, or 3) never written down.
The images show imaginary pages that could have been
written - but they are not perfect final manuscripts. Instead,
these manuscripts are torn, half-erased. They are material
evidence of a cultural process typical of all human cultural
history: the best ideas and plans in all genres of art were
frequently never realized and remained only as notes and
drafts. Or were they were never written down and therefore
simply disappeared completely.
The universality of this history is emphasized by the text we
see in the images. It looks like something that is written in
some existing language but actually it is not: AI simulates
general patterns of grammar, handwriting and hand sketching,
as opposed to any concrete language.
Library of Imaginary Manuscripts: Reading Room
DRAWING ROOMS
Digital images created with generative AI and edited in
Lightroom, 2023.
We see young people talking, smoking, contemplating - but
what interests me are the interior spaces they inhabit, and
accumulation of objects and details in these spaces. Some
objects are placed on the wall-size shelves; others cover
coffee tables, sofas and smaller shelves. In some cases we
can easily identify these objects, but in other cases they are
harder to identify. Some look like fragments, traces and
shadows of the objects that are gone. Yes, these are
“fragments” - but of what?
In ancient art museum we see material fragments of long
gone civilizations. Pieces of vases and plates, metal tools,
jewelry, tiny statues missing limbs or heads. These are real
objects from 3000 or 1500 years ago. But AI “fragments”
have diffident ontology. Generative AI model extracts
structures and patterns from hundreds of millions of images
and distributes them across trillions of connections. In this
process, digital materially of images is further virtualized,
evaporated, diffused.
In generative AI training, digital materially of images is
further virtualized, evaporated, diffused. These are
fragments of fragments. Deposits of already broken forms.
More fragmented then 18th century ruins admired by
painters and visitors in Italy on Grand Tour.
They are not like “glitches.” And they are not noise as
defined in the mathematical theory of communication by
Claude Shannon in the 1940s.
I see the accumulations of objects, their shadows, and traces
in this series as a metaphor for Generative AI process.
Despite their illusionary materiality, the generated
“fragments” you see are like scents, the invisible movements
of air blown by the wind, or tiny broken shells left in the sand
after a wave recedes back into the sea.
The fragments of objects you see in this image series -
breaking from the celling, covering the floor, floating in
space, gathered in strange "clouds" - are ifantoms sent by AI
models into our world. Perhaps in these images AI medium
wants to reveal itself to us - to remind us that the seeming
materiality of synthesized scenes is illusionary and fleeting,
and behind it are endless debris of human cultural history.
CLOSED WORLD
Selected works on paper (drawings, etching), 1981-1985.
My early images frequently depict a "closed world." This
world sometimes has a wall enclosing it in the back, and
other times it has all three walls. In some drawings, the
ceiling serves as an enclosing surface.
These drawings can be interpreted as stage designs for
imaginary plays. You are the observer watching this scene.
But a play - a narrative, a story with actors, motivations, and
actions - never was of interest to me. I was only interested in
scenography. As a result, you rarely see human figures in my
drawings – only space and objects.
Another interpretation: we live in the world which for us is
real, but it is only a performance inside an enclosed theatre
stage. This concept comes from my experience of growing
up in Russia during the late communist era. I experienced
reality around me as a giant surreal performance on the
scale of a whole country. Every day, people were performing
the roles they were assigned in front of one another on a
massive stage.
There is no other outside world. People spend their entire
lives in this theater. There is no redemption, escape, some
“normal” life, or even some small rehearsal area. You get
born into this society and learn its rituals already as a child.
Nevertheless, enclosures like low hanging skies, back walls,
and curtains aren't always perfectly fixed. The corners'
seams came loose over time. To keep them in place, various
pols and nails were added. The simulation is slowly
disintegrating, and perhaps a few actors even managed to
escape. The toilet paper roll is affixed to the rear wall of a
particular area, but we cannot be certain. The typical
construction site objects (we lived next to one such site in
Moscow) progressively became frozen. They became part
the utopia that turned into its opposite. We can see from the
signs that this has been ongoing for at least 800 years and
that construction will never be completed.
2017
THE PROCESS
My digital images are created using a variety of AI tools. Some
of these tools are new, while others have been available to
artists, photographers, and designers for many years.
First, I fed text and multiple visual references into the
Midjourney generative app. These references could include
previously generated images, photos of my early artworks,
paintings, etchings, and engravings from art history, and so
on. The text portions of the prompts are also frequently
purposefully ambiguous, resulting in what I call "controlled
misunderstanding."
I only select a tiny proportion of generated images, similar to
how photographers choose only a few photographs to include
in their portfolio from thousands they shoot and edit. I edit
selected images in the Lightroom app, returning to the same
images many times over the course of months. Using its own
AI, Lightroom can suggest possible edits to me. I also
frequently use its other AI features to automatically select
subjects and objects in images so that they can be edited
separately.
Finally, I use Gigapixel AI to prepare high resolution versions
of images that can be printed at large sizes. This tool predicts
how my images' details will appear at larger sizes .
In the end, my image creation process is a combination of
intuition and logic, drawing on my 50-year experience in the
visual arts (I began art training as a teenager in 1973).
BIOGRAPHY
Lev Manovich is an artist, author, and one of the world's most
influential digital culture theorists. (ChatGPT lists him in the
first place among "the most important thinkers" in the world in
digital art, new media art, new media studies, and digital
culture.) After studying visual art, architecture, and
filmmaking, Manovich began using computers to create digital
art in 1984. His work has been exhibited in 14 solo and 122
international group exhibitions at many prestigious
institutions, such as the Institute for Contemporary Art
(London), Centre Pompidou, The Shanghai Biennale, and The
ZKM | Center for Art and Media.
Manovich played a key role in creating four new research
fields: new media studies (1991-), software studies (2001-),
cultural analytics (2007-) and AI aesthetics (2018-). He is
Presidential Professor of Computer Science at City University
of New York's Graduate Center and the Director of the
Cultural Analytics Lab. His books include Artificial Aesthetics,
Cultural Analytics, Instagram and Contemporary Image, and
The Language of New Media, which has been called "the most
provocative and comprehensive media history since Marshall
McLuhan." Manovich was included in the lists "25 People
Shaping the Future of Design" (Complex, 2013) and "50 Most
Interesting People Building the Future" (Verge, 2014).