Neural networks in processing textual objects of Russian literature

This article explores the capabilities and limitations of neural networks in processing textual objects of Russian literature within educational contexts. The study material comprises responses generated by ChatGPT-4o to questions based on a university-level Russian literature curriculum for philology students. The analysis focuses on three key aspects: interpretation of V. Shukshin’s short story “Srezal”, analysis of A. Solzhenitsyn’s “One Day in the Life of Ivan Denisovich”, and classification of trends in 20th-century Russian poetry. The results indicate that while ChatGPT produces structurally coherent answers mimicking academic discourse, their quality varies by task type. The most significant discrepancies with original texts occur in literary analysis: the AI introduces factual errors (misnamed characters, distorted plot details) and oversimplifies narrative conflicts. Conversely, its accuracy improves when listing historical-literary facts (poetic movements, authors’ names). The study underscores that integrating AI into literary education necessitates robust verification protocols. The article proposes a framework for critical engagement with neural networks, emphasizing source cross-referencing and the cultivation of students’ analytical competencies.