Since OpenAI released the ChatGPT (Chat Generative Pre-Trained Transformer) in November 2022, it has brought the machine translation into a new development mode. However, the gap between the generated translation style and that of the original text still exists, which is an important issue that cannot be avoided in natural language processing (NLP). Taking the Chinese translations of Virginia Woolf’s short essay “How Should One Read a Book?” as the research subject, this paper focuses on three style generation elements: formality, literariness, and classicality. It compares, analyses, and comments on the reasons for the disparity in styles between human translation and ChatGPT translation in order to provide prospects for the development of translation technology in the era of large language models (LLMs).