Scholarlink

<p>Current generative AI educational applications have primarily focused on conceptual interactions, leaving a gap in understanding how generative AI can provide feedback to enhance science learning with science simulations. This study thus proposed a dynamic prompt framework for designing pedagogical agents that addresses the focusing principle to support critical observation and concept development using science simulations. The results indicate that students receiving feedback from the generative AI exhibited greater behavioral engagement compared to those who did not receive feedback. Furthermore, the experimental group achieved higher levels of achievements in the learning session and in the conceptual posttest, indicating a deeper understanding of scientific concepts. Our findings extend the application of generative AI beyond conceptual interactions to enhance science learning with science simulations. The feedback generated by AI within this framework transforms solitary learning into a reciprocal learning process. The affordance and limitations of the generative AI in supporting experiential learning with science simulations are discussed.</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

Enhancing Science Simulation-Based Learning with Dialogic Feedback from Generative AI Agents

  • 期刊论文

  • 发表日期：

    2026年7月31日

查看全文
  • Cai-Ting Wen,
  • Yi-Syuan Chu,
  • Chih-Chung Hsu,
  • Po-Yao Chao,
  • Ying-Siou Chen,
  • Chia Hui Chang,
  • Fu-Kwun Hwang,
  • Chen-Chung Liu

摘要

Current generative AI educational applications have primarily focused on conceptual interactions, leaving a gap in understanding how generative AI can provide feedback to enhance science learning with science simulations. This study thus proposed a dynamic prompt framework for designing pedagogical agents that addresses the focusing principle to support critical observation and concept development using science simulations. The results indicate that students receiving feedback from the generative AI exhibited greater behavioral engagement compared to those who did not receive feedback. Furthermore, the experimental group achieved higher levels of achievements in the learning session and in the conceptual posttest, indicating a deeper understanding of scientific concepts. Our findings extend the application of generative AI beyond conceptual interactions to enhance science learning with science simulations. The feedback generated by AI within this framework transforms solitary learning into a reciprocal learning process. The affordance and limitations of the generative AI in supporting experiential learning with science simulations are discussed.