Art match-mashups

_Convergence of art that meets AI and machine that learns art

Youngeun Museum of Contemporary Art

2021. 11. 6 – 2021. 11. 28

Artists: Kim Changkyum, Yang Daewon, Yoo Hanyi, Lee Donsoon, Hong Kyoungtack

AI technical director: Cho Hyungrae(Researcher of Intelligence Media Research Center at Seoul National University of Science and Technology)

AI SW Development: Ko Youngchan, Jang Junyoung, Jang Ilsik, Jang Chanho (Researcher of Intelligence Media Research Center at Seoul National University of Science and Technology)

AI Technical advice: Park Kuman(Professor of information and communication engineering at Seoul National University of Science and Technology)

Pre-Processing of AI datasets Ⅰ: Lee Narae · Ahn Siin · Choi Narin · Choi Jeongwon

 Pre-Processing of AI datasets Ⅱ: Gachon University of art

(Kang Doyeon · Kim Wooseop · Kim Yujin · Park Gyutaek · Bang Hyeonho · Jeong Gayeon · Joo Jaehong)

Design: Lee Narae

Project planning: open space BLOCK’s

Sponsorship: Youngeun Museum of Contemporary Art

pecuniary aid: Art Council Korea(ARKO)

“We shape our tools and thereafter our tools shape us”. This quote by Marshall McLuhan speaks more loudly now than ever before, as technology continues to advance at a dazzling pace. In a metaphorical sense, we may have already become cyborgs trapped in a network of capital. Ever since the advent of the age of mechanical reproduction, our ability to adapt to new technologies has not only increased efficiency but constantly reshaped the sensibility of the times and the foundation of our lives. In particular, as the entire world is going through the unprecedented COVID-19 pandemic, we are quickly transitioning to a non-contact society and a new digital online-based paradigm, the metaverse. Against this backdrop, the Youngeun Museum of Contemporary Art is hosting a special exhibition in Exhibition Hall 1 featuring the outcomes of a research project in two formats: a demonstration of the research process and an exhibition. Five visual artists and five researchers of the Intelligent Media Research Center of Seoul National University of Science and Technology have worked together on a joint project on the theme of artificial intelligence.

The exhibition was planned by open space BLOCK’s and sponsored by Art Council Korea(ARKO) as part of a project to bring art and technology together. The exhibition represents an experiment in creating art according to a dialectic approach whereby AI massively learns the different styles of the five featured artists, namely, Kim Changkyum, Yang Daewon, Yoo Hanyi, Lee Donsoon and Hong Kyoungtack, and creates new works of art (products) based on the descriptions provided by the five artists. Then, the artists reinterpret the AI-created outcomes or products by incorporating their own unique perspectives in order to come up with their final works. In the meantime, four researchers, Jang Ilsik, Ko Youngchan, Jang Junyoung, and Jang Chanho, of the Intelligence Media Research Center formed a team supervised by professor Park Kuman and led by artist Cho Hyungrae, the technical director of the project. The team went through repeated machine learning of the five artists’ styles and descriptions, improving the performance of the AI-based technology - including StyleGAN2-ada and Pix2pix HD - in the process and exploring the possibility of merging artistic data and AI.

AI, which is deeply and widely embedded in our lives, is known for its ability to perform multiple jobs in art and creative activities, through machine learning involving big data, resulting in convenience, playfulness, creativity, changeability via technological manipulation, and increased labor efficiency in the art production process. Particularly, AlphaGo has made it common knowledge that AI can surpass humans in certain fields. In this context, it remains an intriguing question whether or not AI can mimic certain artists through machine learning and ultimately create art that can fool people’s eyes, based on the descriptions of the artists. In other words, can AI create an optical illusion in painting that can replace conventional art by captivating the viewers and touching their hearts?

The exhibition title “Art Match-Mashups” succinctly summarizes the significance of the convergence project, i.e. comparing and analyzing original works of art created by the artists based on the common thread of visual art, and the outcomes produced by AI that has learned the techniques of the same artists, in different phases, and ultimately testing the possibility of creating art through collaboration between humans and AI. Like a mashup that takes content from different multiple sources and blends them to create new songs, videos, or computer files, this project aims to freely cross the boundaries of humans and AI and develop new forms of creative work by exchanging, integrating and applying the artists’ unique creative data (content) and technology. In the process, AI learns the artists’ unique techniques and uses the descriptions of their works as big data, creating artificial art in the form of a collage. Then, the artists come up with new ideas and motifs and create their final works, which include the collective learning process inherent to them. This dialectic experiment in creating art is what the exhibition is all about. The exhibition also provides a forum where people can come and appreciate the joint creative works of AI and the artists.

This is not the first project of its kind to bring art and technology together. However, previous attempts focused on merely comparing the technical performances of machines and humans. This project is unprecedented and unique in that AI learns the unique styles of particular artists and their descriptions of their own works and then produces various outcomes, after which the artists reinterpret the AI-created works and create their final works for presentation and evaluation in an exhibition in a unique style that combines both the process and the outcomes and features academic research and artistic creation. In this sense, the exhibition, which is the first research project to present the possibility of collaboration between practicing artists and AI in a dialectic approach, includes an active process in which the artists share their long-accumulated creative experience with AI and embrace AI as a source of support, thereby finding common ground and identifying differences between the two in the collaborative process of learning and creation. This indicates a step forward in the evolution of human creative experience. In other words, while research in visual art had become bogged down in the phase of AI machine learning and the production of results, this project actually attempts to show that the barriers between AI and humans can be brought down and creative works can be taken to a new level by blending human experience and thinking processes into the process of creating art.

Art Match-Mashups will show you a large quantity of images that AI has created by learning the unique styles of the five artists, which will help you to understand the different phases of AI’s technological evolution. In addition, the project will be a unique opportunity to glimpse the possibilities that can be unleashed when technology and humans work together and when AI and art meet, including creation, artistic changeability, humans’ formative ability, and irreplaceable perspectives.

The exhibition adopts the unusual format of displaying, along the visitors’ movement path, constructing learning data sets, the production flow of the artists’ new works, their existing works, the statements of the artists, the final AI-created images, the final works of the artists, a large-scale visual artwork that combines the data AI has learned and the outcomes, AI interpolation, 100 AI-created items to be used in an audience survey and preference voting, and an archive of large quantities of filed AI-created items. The content of the project on display will be compiled as part of a research project and the findings will be published as an academic thesis. The project will conclude with a final survey of the artists and the AI-created outcomes during the exhibition period.