Mutual Reinforcement of Predictions and Artistic Illusions in Generative Language Modeling

Artificial intelligence systems, represented by ChatGPT, are revolutionizing society and art with a speed and density never seen before. Therefore, when generative language models intervene in art, constructing contexts and contexts of human consciousness and behavior, and creating simulations and generating "predictive" information that reinforces the artist's creative intent, the perception of art's virtual image will be enhanced. The anchoring of data and information selected under the guidance of the artist's subjectivity and the feedback constructed by the Internet crawler system carries a projection of the artist's own. The commonality between consciousnesses is in fact the reading of "illusion". In this paper, we take the relationship between the "prediction" of generative language modeling and artistic illusion as a research content, and explore the unconscious artistic illusion produced when facing the predictive "information". It tries to explore the possibility of creative art practice that grows organically between artificial intelligence and art.