Scholarlink

The purpose of this study was to evaluate the effectiveness and relevance of AI-powered non-player characters (NPCs) in educational settings within the metaverse platform Classlet, particularly within the domain of applied social sciences. Integrating Speech Act Theory, Prospect Theory, and the Recognition-Primed Decision Model into AI prompt templates enhanced interactions and positively impacted student engagement. The AI NPCs achieved a performance ratio of 128%, providing contextually relevant responses effectively maintained by the prompt templates. A survey on students’ perceptions of the AI and VR integration revealed positive user perceptions, with strong correlations between enjoyment, perceived usefulness, and intent to use. Correlational analysis showed a strong fit (R2 = 0.816) for user intent, which had a mean of 74%. U-tests also indicated that female and non-VR users encountered more technical difficulties. Overall, the results suggest new ways of using AI and VR to promote learning engagement and interactions within applied social sciences.

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

AI NPCs in an Educational Metaverse: Evaluating the Effectiveness of Prompt Templates for Contextual Interactions

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Wilkinson Daniel Wong Gonzales,
  • Daniel Jiandong Shen,
  • Aihua Yan,
  • Nina Xie,
  • Maria Leonora Francisco,
  • Paulina Pui Yun Wong

摘要

The purpose of this study was to evaluate the effectiveness and relevance of AI-powered non-player characters (NPCs) in educational settings within the metaverse platform Classlet, particularly within the domain of applied social sciences. Integrating Speech Act Theory, Prospect Theory, and the Recognition-Primed Decision Model into AI prompt templates enhanced interactions and positively impacted student engagement. The AI NPCs achieved a performance ratio of 128%, providing contextually relevant responses effectively maintained by the prompt templates. A survey on students’ perceptions of the AI and VR integration revealed positive user perceptions, with strong correlations between enjoyment, perceived usefulness, and intent to use. Correlational analysis showed a strong fit (R2 = 0.816) for user intent, which had a mean of 74%. U-tests also indicated that female and non-VR users encountered more technical difficulties. Overall, the results suggest new ways of using AI and VR to promote learning engagement and interactions within applied social sciences.