期刊文献+

教育提示语工程:构建数智时代的认识论新话语 被引量:4

Educational Prompt Engineering: Constructing a New EpistemologicalDiscourse in the Era of Digital Intelligence
原文传递
导出
摘要 人类在漫长的发展历程中经历了“日常语言-符号语言-数字语言”的话语转向,每一次转向都改变了人们对世界的看法或与世界互动的方式。教育提示语作为数智时代的认识论新话语,在大语言模型的教育应用中提供加速力量,以因人而异的对话内容、循环连贯的对话进程、高质稳健的对话产出为学习者创造定制化学习体验,为教育者提供全流程设计支持,为管理者带来针对性决策见解。为了释放大语言模型在教育中的潜力,文章提出了教育提示语设计的“CORE”框架,该框架由必备要素(语境、目标)和可选要素(角色/规则、示例)构成,归纳出基础提示、示例提示、角色提示、规则提示、组合提示五种提示语类,提供了结构化提示模板设计范例。文章最后建议,实现教育提示语工程需要创建面向不同场景的教育提示库,研制教育提示语工程的技术规范,培养教育领域的“提示语设计师”,测评提示语表现性能并进行风险防范,以期为新一代人工智能赋能教育变革注入新活力。 In the long process of human development,there has been a discourse shift from everyday language,to symbolic language,and then to digital language.Every turn has changed people's perceptions of the world or the way they interact with it.Educational prompts,as a new epistemological discourse in the era of digital intelligence,provide acceleration power in the educational application of large language models,and create customized learning experiences created for learners with personalized dialogue content,continuous and coherent dialogue processes,and high-quality and stable dialogue outputs,which provides full-process design support for educators and brings targeted decision-making insights for administrators.To unleash the potential of large language models in education,this paper proposes a"CORE"framework for designing educational prompts,consisting of essential elements(Context and Objectives)and optional elements(Role/Rule and Example).The paper categorizes five types of prompts:foundational prompts,example prompts,role prompts,rule prompts,and composite prompts.In conclusion,it is suggested that to realize educational prompt engineering,it is necessary to establish educational prompt libraries tailored for various scenarios,develop technical specifications for educational prompt engineering,nurture"prompt designers"within the educational domain,evaluate prompt performance,and implement risk mitigation measures.This is expected to inject new vitality into the new generation of AI-empowered education transformation.
作者 赵晓伟 祝智庭 沈书生 Xiaowei Zhao;Zhiting Zhu;Shusheng Shen
出处 《中国远程教育》 2023年第11期22-31,共10页 Chinese Journal of Distance Education
基金 2023年度国家社会科学基金重大项目“新一代人工智能对教育的影响研究”子课题“应对新一代人工智能挑战的教育管理改革和政策创新”(项目编号:VGA230012)的研究成果。
关键词 教育提示语工程 生成式人工智能 大语言模型 ChatGPT 人机协同 educational prompt engineering generative artificial intelligence large language model ChatGPT human-machine synergistic
  • 相关文献

参考文献4

二级参考文献53

共引文献83

同被引文献111

引证文献4

二级引证文献3

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部