Credit: Kikuyama
So Kanno (JP), Takahiro Yamaguchi (JP)
Characters are a means of visual communication and recording a language. Civilizations throughout the world have created various characters that convey their culture and history. This project focuses purely on the form of the characters rather than their meaning.
The characters have been learned by artificial intelligence (AI) not for their meaning but for their shape and patterns. The AI has created and drawn lines that look like characters but do not have any meaning.
This work was publicized at the Aichi Triennale 2016 international art festival. It was implemented by collecting handwritten artist statements or descriptions of work by an extremely broad international group of ten participating artists. By learning handwriting with one writer in each language, the artificial intelligence collected information on the shapes of each character system, as well the idiosyncrasies of each writer. The lines generated are written as if they mean something important; they also look deceptive.
Credits
Machine Learning Programming: Hironori Sakamoto
Supported by Nihon Unisys, Ltd., Haps
Sponsored by Japan Media Arts Festival and Bunkacho – Agency for Cultural Affairs, Government of Japan
Handwriting provided by: Valsan Koorma Kolleri, Lai Chih-sheng, Gulnara Kasmalieva & Muratbek Djumaliev, Kio Griffith, Ali Cherri, Taloi Havini, Song Sanghee, Shreyas Karle, Kawayan De Guia, Uudam Tran Nguyen
