Potencjał i ryzyko stosowania dużych modeli językowych w przekładzie chińskich tekstów prawnych
Assessing the Potential and Limitations of Large Language Models in the Translation of Chinese Legal Texts into Polish

Summary/Abstract: The article discusses the potential and limitations of the Large Language Models (LLM) in translation of Chinese legal texts into Polish. The theoretical part covers the specifics of legal Chinese, including its historical background and systemic constraints, and the rising significance of the translatory abilities of the LLMs. The next section contains the data regarding the results of empirical research on linguistic abilities of the language models on the example of the SLING benchmark that depict significant discrepancies between language processing of the language models and precision of human assessment. The analytic part compares translations generated by particular commercial language models and illustrates various translatory techniques that result in semantic and normative shifts. The conclusion supports an idea that the LLMs should only be used as auxiliary tools in the process of legal translation, requiring constant expert supervision.

• Details
• Contents