Art text generator, interpretations, 30 works

CAFA Art Museum, Beijing

2021

    Tong crawled a large number of art texts from the Internet and used this as a corpus to train an AI (artificial intelligence)* so that it can automatically generate art texts. He submit to the generated text as interpretations of artworks, which don't exist yet, to make works. That is, first obtain the interpretations of the work in the usual sense, and then make the work, basing on these random paragraphs.

​

    * The AI is based on JPT2-Chinese, and the text data comes from: art portals, art topic texts in online forum communities, art WeChat public accounts, search engine art terms search results, etc., a total of 250,303,623 bytes.