Scholarlink

<p>Science education research that draws on machine learning (ML) methods represents a&#xa0;rapidly and dynamically developing field. Against the background of much work focusing on assessment or evaluations of ML models for their validity and effectiveness, this paper argues that future research should place greater emphasis on the transformative potential of ML for teaching and learning without losing sight of ethical and epistemological challenges as part of critical reflection. Using two continuums spanned by the antipodes of “basic research and practical orientation” and “incremental and disruptive innovation,” these perspectives are brought together through four exemplary research areas: First, “Individualizing instruction”; second, “Understanding learning processes through physiological sensors and multimodal analysis”; third, “Integrating qualitative and quantitative data”; and finally, “Researching with AI”. This paper uses the state of international research and science didactic problems to further specify the potential of ML for science education. Orientation is suggested to researchers, and key challenges for the advancement of the field are described that could inform science education research in ML in the coming years.</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

Künstliche Intelligenz in den Naturwissenschaftsdidaktiken – gekommen, um zu bleiben: Potenziale, Desiderata, Herausforderungen

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年2月20日

查看全文
  • Andreas Nehring,
  • David Buschhüter,
  • Marcus Kubsch,
  • Tobias Ludwig,
  • Peter Wulff,
  • Knut Neumann

摘要

Science education research that draws on machine learning (ML) methods represents a rapidly and dynamically developing field. Against the background of much work focusing on assessment or evaluations of ML models for their validity and effectiveness, this paper argues that future research should place greater emphasis on the transformative potential of ML for teaching and learning without losing sight of ethical and epistemological challenges as part of critical reflection. Using two continuums spanned by the antipodes of “basic research and practical orientation” and “incremental and disruptive innovation,” these perspectives are brought together through four exemplary research areas: First, “Individualizing instruction”; second, “Understanding learning processes through physiological sensors and multimodal analysis”; third, “Integrating qualitative and quantitative data”; and finally, “Researching with AI”. This paper uses the state of international research and science didactic problems to further specify the potential of ML for science education. Orientation is suggested to researchers, and key challenges for the advancement of the field are described that could inform science education research in ML in the coming years.