Scholarlink

In this paper, we explore the integration of large language models (LLMs) in interactive digital narratives (IDNs) in the context of remediating the classic Chinese novel, Hong Lou Meng (HLM). Adopting Research through Design (RtD), we developed an LLM-driven IDN prototype that allows players to interact with Lin Daiyu, the iconic female character of HLM. We then playtested the prototype with seven HLM readers and five IDN designers through semi-structured interviews. Our findings reveal the potential and challenges of using LLMs as character and plot design materials for IDNs that remediate classic literature. We conclude by proposing two design strategies for LLM-driven IDNs that ensure both player engagement and narrative coherence: 1) Designing towards engaging AI and humans in the same task, and 2) Expanding AI responses beyond predictable boundaries.

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

Meet Daiyu in The Era of GenAI: Using LLMs to Remediate Classics into Interactive Digital Narratives

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Yuxuan Huang,
  • Kexin Xiang,
  • Gauransh Sharma,
  • Reshad Rabbi,
  • Ran Ju,
  • Jussi Pekka Holopainen

摘要

In this paper, we explore the integration of large language models (LLMs) in interactive digital narratives (IDNs) in the context of remediating the classic Chinese novel, Hong Lou Meng (HLM). Adopting Research through Design (RtD), we developed an LLM-driven IDN prototype that allows players to interact with Lin Daiyu, the iconic female character of HLM. We then playtested the prototype with seven HLM readers and five IDN designers through semi-structured interviews. Our findings reveal the potential and challenges of using LLMs as character and plot design materials for IDNs that remediate classic literature. We conclude by proposing two design strategies for LLM-driven IDNs that ensure both player engagement and narrative coherence: 1) Designing towards engaging AI and humans in the same task, and 2) Expanding AI responses beyond predictable boundaries.