AI 와예술 : 이론적배경과최근동향 김태희영산대학교교수 ( 전 ) AIRI 책임연구원 2018.3.14
목차 Introduction 현대의예술과미디어아트인공지능에서표현에관한문제딥러닝과예술 Discussion
AI 와예술의관계 AI 는예술의도구 예술작품은도구의성격을내려받는다. AI 를통한예술개념의확장 현대의예술은 개념 도예술의소재 생성, process, time, space, life, media, interaction, authorship 전통예술의연장선에서 로봇이그림을그리고 딥러닝이하는회화 로봇은움직이는조각품 예술비평의관점에서 What, How 3
예술에서도구의영향 원시인의손바닥 vs. 붓 유화 vs. 수채화 대리석 vs. 천 eg., a revolution: tin tubes uniform color and mobility 예술은기술의영향을받는다 4
현대의예술과미디어아트
모나리자 http://www.turkishny.com/news/mona-lisa-yine-evinde-kald
아트의전환기 Marcel Duchamp Fountain, 1917 Photograph by Alfred Stieglitz, 1917 - Wikipedia
Duchamp the creative act is not performed by the artist alone; the spectator brings the work in contact with the external world by deciphering and interpreting its inner qualifications and thus adds his contribution to the creative act. 창의성의발생시키는것은예술가만하는것이아니다 ; 관객또한작품속에담긴뜻을풀어내고해석해 냄으로써작품이세계와더욱폭넓게만나게하는 창조적기여를하는것이다. ( 관객의경험이작품 ) The Creative Act Marcel Duchamp, in The Writings of Marcel Duchamp, ed. by M. Sanouillet and E. Petersonp., p.140, Da Kapo Press, 1973.
[T]here has never been, never will be a unique word, a master-name. 그어디에도유일한, 대표성을가지는이름은없다. Jacques Derrida, Margines of Philosophy. Translated by Alan Bass (Chicago: The University of Chicago Press, 1982), 27.
world picture, when understood essentially, does not mean a picture of the world but the world conceived and grasped as picture. 세상을본다는것은그세상을근본적으로본다고하기보다는사진을 찍듯이자신의눈에비춰지는것만보는것이다. Martin Heidegger, The Question Concerning Technology, and Other Essays. Translated by William Lovitt (New York: Harper Perennial, 1982), 129
a window to see through in which medium is the message 미디엄이메시지인것과같이창문을통하여쳐다보는것 Anne Friedberg, The Virtual Window (Cambridge, Mass.: MIT Press, 2006), ch. 1.
For Duchamp, ate is never the same is eat. - Marjorie Perloff, original emphasis, 2001
John Cage, 4 33, 1952 https://www.youtube.com/watch?v=gn2zclbr_vm
Each something is a celebration of nothing that supports it. - John Cage, 1973 色卽是空 空卽是色 - 반야심경
N009_ 로봇 K-456, 백남준, 1964 http://cyberneticzoo.com/robots-in-art/1964-robot-k-456-nam-june-paik-korean-shuya-abe-japanese/
관객에게반응하는작품 each Beast, 테오얀슨, 2005 The Senster, 1970 https://www.atariarchives.org/artist/sec10.php
자아의확장개념 Exoskeleton by Stelarc, 2003 https://www.youtube.com/watch?v=4vhas9uwfhy
Theo Jansen, 2005 https://www.youtube.com/watch?v=2o9phl3uzoy
Camille Utterback - Untitled 5 https://www.youtube.com/watch?v=zoydp3dxwrk
Daniel Rozin, 1999 Interactive art with wooden mirrors - The wooden mirror, Daniel Rozin, 1999 https://www.youtube.com/watch?v=bzysu9qccem
Seen (2002) video installation by David Rokeby https://www.youtube.com/watch?v=0ai49w7qppy
Karl Sims 1994 https://www.youtube.com/watch?v=jbgg_vsp7f8
It s way of Looking Taehee Kim
미디어아트의개념 관계 / 상대성 (relation/relativity) 변화 (change) 창발 (emergence) 생성 (generation) 거울 (mirror) 장소성 (situatedness) 시간계열변화 (timeline change) 다양한차원의관점 (multiperspectives) 1 차개념 가상 (virtuality) 2 차개념 3 차개념 Subjectivity Identity Agency
인공지능에서표현에관한문제 Representation
로봇을위한명령 : 사과를집어서접시에올려라. 사과 를찾으면 접시에올린다. 만약 ~ 하면 ~ 한다. If ~ Then ~ 26
로봇이사과를어떻게알아보는가? 사과 공 오렌지 복숭아 사과 오렌지 공 복숭아 사과 27
그런데, 세상에는참다양한사과가있다. 사과는사과인데, 어떤사과? 우리는어떻게사과를사과로아는가? 우리각자가알고있는사과는정말같은사과인가? 28
Symbol Grounding Problem How words(symbols) get their meanings and hence to the problem of what meaning itself really is.
의자, 잎사귀 나 게임 과같은말들을 애매하지않게그리고논쟁거리가되지않게적용하려면 우리는무엇을알아야하는가? - Wittgenstein
道可道非常道 名可名非常名 - 노자
Deep Learning and Art
신경망의구조예 http://www.kdnuggets.com/2016/01/seven-steps-deep-learning.html 딥러닝은깊은신경망 신경망은정보를나름의방식으로저장, 분류, 학습
Inceptionism Google 의소프트웨어엔지니어들이공개한 학습된신경망이어떻게이미지를바라보는 지, 사물이나동물등을어떻게이해하며생각하는지를보여주는실험 (2015.7) 랜덤이미지에서출발하여바나나그리기 : 신경망이알고있는바나나 https://research.googleblog.com/2015/06/inceptionism-going-deeper-into-neural.html
Inceptionism 이미지를제시하면신경망이하는해석의결과를볼수있다. Low Level: 모서리나모서리의 orientation 등에반응 선, 작은패턴으로구성 https://research.googleblog.com/2015/06/inceptionism-going-deeper-into-neural.html
Inceptionism 이미지를제시하면신경망이하는해석의결과를볼수있다. High Level: 더욱구상화된패턴, 사물의모습등으로표현 https://research.googleblog.com/2015/06/inceptionism-going-deeper-into-neural.html
Gatys, Leon A., Alexander S. Ecker, and Matthias Bethge. "A neural algorithm of artistic style." arxiv preprint arxiv:1508.06576 (2015).
Convolutional Neural Network http://www.kdnuggets.com/2016/01/seven-steps-deep-learning.html
작가 : 진코건 (Gene Kogan, b.1985, 미국 ) A Book from the Sky created by Deep Convolutional Generative Adversarial Network(DCGAN) trained on database of handwritten Chinese characters http://genekogan.com
작가 : 진코건 (Gene Kogan, b.1985, 미국 ) http://genekogan.com
작가 : 하싯아그라왈 (Harshit Agrawal, b.1992, 인도 ) http://harshitagrawal.com/works/tandem 41
인간과 AI 의 Collaboration 무용수가춤을춘다 Louise Crnkovic-Friis Dinner with Friends http://crnkovic-friis.com/work/dinner-with-friends/ Deep Neural Network 은자신이보고있다고생각하는것을보여준다 무용수는그것을보고새로운안무를창안한다 반복. 42
Embodiment & Situatedness
Designing a Robot 광원 Vehicles, Braitenberg (1983)
Designing a Robot 광원 Vehicles, Braitenberg (1983)
Intelligence is determined by the dynamics of the interaction with the world. 지능은세계와의상호작용에의하여정해진다. - Rodney Brooks (1991) Structural Coupling Francisco Varela (1992) : A well managed relationship reduces the amount of explicit information processing 개체와환경이조화롭게관계가만들어지면계산해야할양도줄어든다. - my Interpretation
이작품처럼, 놓여있는오브제만이작품이라할수없으며 어떤특정한장소에놓여졌다는것이자신을구성하는요소인것과 이를경험하는모든사람들의경험을통틀어서작품이라할수있는것 Tiled Arc, Richard Serra (1981)
Discussion Representation( 표현, 표상 ) 의다양성 지능은현상일것이다. 몸과마음은분리할수없다. 객체와의미는분리할수없다. 행동 ( 또는기능 ) 은형태 ( 조형 ) 와호환된다. ( 그렇다면의미와물질은호환가능한가? 물질은현상이므로?) 생각은 form 인가? 경험은 form 인가? (memory 는?) 그리고관계 (relationship) 의문제 인공지능은철학에게서가져올것이많다. 과학과예술의연결은실천적인철학을통하여이루어지길