Scholarlink

<p>In this article, we examine the usefulness of artificial intelligence for literary studies research by focusing on two typical activities in the field. To this end, two sets of experiments were conducted using large language models to produce annotations for a&#xa0;literary text and to generate the text of three subsections of the present article. In the first section, summaries of scientific articles on LLMs in literary studies were generated. In the second section, ChatGPT was prompted to generate a&#xa0;description of our annotation experiments based on a&#xa0;predefined structure, while in the third section the results of the annotation experiments were transferred to ChatGPT for analysis. The design of these experiments takes into account the technical skills typically possessed by literary scholars. The writing experiments were conducted using ChatGPT via the chat interface in the browser, while Llama 3.1 was used locally for the annotations, which requires at least basic programming skills. Although both sets of experiments were partially successful, the current state of these language models does not yet seem suitable for achieving satisfactory results within a&#xa0;reasonable timeframe with limited resources.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Literaturwissenschaftlich Arbeiten mit großen Sprachmodellen?

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年5月15日

查看全文
  • Mari Akazawa,
  • Evelyn Gius

摘要

In this article, we examine the usefulness of artificial intelligence for literary studies research by focusing on two typical activities in the field. To this end, two sets of experiments were conducted using large language models to produce annotations for a literary text and to generate the text of three subsections of the present article. In the first section, summaries of scientific articles on LLMs in literary studies were generated. In the second section, ChatGPT was prompted to generate a description of our annotation experiments based on a predefined structure, while in the third section the results of the annotation experiments were transferred to ChatGPT for analysis. The design of these experiments takes into account the technical skills typically possessed by literary scholars. The writing experiments were conducted using ChatGPT via the chat interface in the browser, while Llama 3.1 was used locally for the annotations, which requires at least basic programming skills. Although both sets of experiments were partially successful, the current state of these language models does not yet seem suitable for achieving satisfactory results within a reasonable timeframe with limited resources.