Neural Kimono, or Intertextuality of AI

Semiotician Julia Kristeva wrote in her essay "Word, Dialogue and Novel" [PDF] about texts, they were "absorption and transformation of another [texts]". She popularized the term "Intertextuality", which is There are no original texts; they are based on different previous works by continuously metamorphosing our narratives. Texts are textures -- they are material of our modern writing-based storytelling with its patterns, topics, ideas, and indeed have other features of everlasting change compared to everchanging stories, told over generations in Ancient times (in the way of "Chinese Whisper" game, with changing contents during timely transformed contexts). With the post-authorial storytelling in the times of AI, we have to rethink intertextuality as a generative driver. If the author-based text was created by the conscious and subconscious human body of thoughts (author's knowledge, experience, biases, stereotypes, preoccupations, distorted memories), Artificial Intelligence as creator is trained on previous datasets and generates new works by re-interpreting, re-combining known features, patterns, perspectives (like in self-attention based language model GPT-3).