华东师范大学学报(教育科学版) ›› 2023, Vol. 41 ›› Issue (7): 91-102.doi: 10.16382/j.cnki.1000-5560.2023.07.009

• 学科变革与学科建设 • 上一篇    下一篇

计算教育学视域下的ChatGPT:内涵、主题、反思与挑战

郑永和1, 周丹华1, 张永和2, 田雪葳3, 王晶莹1, 郑一4   

  1. 1. 北京师范大学科学教育研究院, 北京 100875
    2. 深圳大学教育学部, 深圳 518060
    3. 青岛大学师范学院, 青岛 266071
    4. 民航总医院, 北京 100123
  • 出版日期:2023-07-01 发布日期:2023-06-25
  • 基金资助:
    国家自然科学基金项目“学习环境对中学生全球素养的影响机制与循证决策研究:基于机器学习的关联规则挖掘”(72074031);国家社会科学基金重大项目“深入推进科技体制改革与完善国家科技治理体系研究”(21ZDA017)

ChatGPT from the Perspective of Computational Education: Connotation, Theme, Reflection, and Challenge

Yonghe Zheng1, Danhua Zhou1, Yonghe Zhang2, Xuewei Tian3, Jingying Wang1, Yi Zheng4   

  1. 1. Research Institute of Science Education, Beijing Normal University, Beijing 100875, China
    2. Faculty of Education, Shenzhen University, Shenzhen Guangdong 518060, China
    3. Normal College, Qingdao University, Qingdao Shandong 266071, China
    4. Civil Aviation General Hospital, Beijing 100123, China
  • Online:2023-07-01 Published:2023-06-25

摘要:

以ChatGPT为代表的生成式人工智能技术进阶推动了计算教育时代数据密集型范式的转型升级,并将计算教育学推向发展关键期。本文首先探讨以ChatGPT为代表的生成式人工智能大模型的价值内涵,从而揭示人工智能进阶推动计算教育学范式升级的要旨。通过使用社会网络分析和数据挖掘方法探讨“师-生-机”知识生成所涌现的教育研究主题,并根据核心领域关系图谱从技术突破、学生学习、教师教学和学校教育四个方面解析学校场域中“师-生-机”多主体研究共同推进计算教育学研究的纵深发展,由此勾勒出强算法算力驱动计算教育学迭代的人机共融的多元化研究图景。再进一步反思ChatGPT教育应用的工具而非目的性,ChatGPT作为一种协助写作的技术手段而非负责任的主体;作为教学的增强和补充方式而非取代教师角色;配合辅助学生学习而不可产生过度依赖。最后从理论建构与决策赋能方面探讨ChatGPT带来计算教育学发展的关键挑战,即探索基于教育计算的基础理论建构、推进计算教育学的结构规则演进、践行计算教育学的环境构建与应用实践、提升教师能力达成教育计算的育人取向、防范智能应用数据的隐私与偏见风险。

关键词: 计算教育学, 生成式人工智能, ChatGPT, 教育大数据, 机器学习, 数据挖掘, 高阶思维

Abstract:

The advancement of artificial intelligence technology, represented by ChatGPT, has driven the transformation and upgrading of data-intensive paradigms in the era of computational education, and pushed computational education towards a critical period of development. This article first explores the value connotation of the generative artificial intelligence model represented by ChatGPT, in order to reveal the essence of promoting the upgrading of computational education paradigm through the advancement of artificial intelligence. Secondly, we explore the emerging educational research themes of teacher-student-machine knowledge generation through social network analysis and data mining, and analyze the multi-agent research of teacher student machine in the school field from four aspects: technological breakthroughs, student learning, teacher teaching, and school education based on the core domain relationship graph to jointly promote the in-depth development of computational education research, which outlines a diversified research landscape of human-machine integration driven by strong algorithmic computing power in computational education iterations. We further reflect on ChatGPT’s educational application as a tool rather than the ultimate goal, that is, ChatGPT serves as a technical means to assist writing rather than a responsible subject, as an enhancement and supplement to teaching rather than replacing the role of a teacher, and it is required to cooperate with assisting students in learning and not become overly dependent. Finally, we explore the key challenges that ChatGPT brings to the development of computational education from the perspectives of theoretical construction and decision-making empowerment, namely exploring the basic theoretical construction based on educational computing, promoting the evolution of structural rules in computational education, practicing the environmental construction and application practice of computational education, enhancing teachers’ abilities to achieve educational orientation in educational computing, and preventing privacy and bias risks of intelligent application data.

Key words: computational education, artificial intelligence, ChatGPT, education big data, machine learning, data mining, higher order thinking