Abstract: Artificial Intelligence (AI) has achieved initial success in literary writing, demonstrating potential to compensate for existing shortcomings. However, AI-generated literature lacks a social-historical dimension. The effective connection between the speaking subject, linguistic signs, and the external world constitutes the foundational activity of social-historical existence. Language use has two dimensions:one directed toward the external world and the other toward internal linguistic rules. The application of internal linguistic rules is sufficient for expressing meaning. Although AI constructs sentences based on probabilistic statistics, it lacks the capacity for real-time reporting of external events. The concept of “intertextuality” in textual studies is more appropriate for discussing AI's literary writing. AI writing is a series of texts referencing each other, with no experiential “warmth” behind the texts. AI functions as a literary production tool rather than a historical agent. Both structuralism and AI language models address internal language relations, yet their philosophies are opposite. Structuralism tends to obscure changes caused by social history, while AI's probabilistic statistics retain a channel for historical information, though these statistics can only indirectly demonstrate historical transformations rather than accurately describe them. They cannot reveal the sudden fluctuations of history in the short term. Only human writers can engage in dialogue with history; AI excels only in word statistics, recording, and composition. The increase in AI-generated output is often linked to economic profits, pushing this output toward mass literature. Algorithms and precisely targeted data inputs risk fostering vast information cocoons, challenging the literary concepts underlying the classical literary system. It is crucial to recognize this impending transformation.