摘要
自2022年底以来,大型语言模型ChatGPT以爆炸式用户增长与网络传播的方式向公众宣告其诞生,并因其流畅的多轮次对话、对需求的准确理解、实用性高的文本(含程序代码)生成,受到教育领域的广泛关注。对于以ChatGPT为代表的大型语言模型及其衍生应用的教育技术学思考,不能停留在为教师和学生提供使用建议的层面,还需更深入。其中的一个方向是探讨源自教学机器的适应性教学系统及其所支持的个性化学习对学校教学系统结构的可能影响。对ChatGPT教学行为的分析表明:在教学行为(活动、任务)层面,它能够进行对话辅导、技能训练、讲授演示、协同创作、交互评价等活动,但是缺少课时、单元、课程等上层结构,而作为其行为基础的“动作”,也面临可信性与准确性不高的挑战。尽管如此,ChatGPT表现出来的跨领域自然语言理解与生成能力仍为智能教学系统提供了新型“能力基座”,有望大范围改变智能教学系统的开发模式,提升其教学表现,为个性化学习提供更全面、更有效的支撑。学校的教学系统是否会因此产生结构性变革?本文将“会不会”问题,分为“能不能”与“应该不应该”两个问题,分别从事实逻辑层面、实践价值层面进行探讨。分析表明,学校现有的分科课程、班级授课、教师主导等系统结构仍保持稳定,并未受已有的智能教学系统与慕课等资源太大的冲击;基于大型语言模型的教学自动化还需要补足“上层结构”,夯实“基础动作”;对于基础教育,学校教学系统的要素(如教师)可能会发生很大变化,但系统结构很可能会保持稳定。
Since the end of 2022,ChatGPT,a Large Language Model,has announced its birth to the public with explosive user growth and online dissemination.Due to its smooth multi-round dialogue,accurate understand-ing of requirements,and high practical text generation including program coding,it has attracted wide attention in the field of education.The educational technology thinking of the Large Language Model represented by ChatGPT and its derivative applications should not stop at the level of providing suggestions for teachers and students,but should go deeper.One direction is to explore the possible impact of Adaptive Teaching Systems derived from Teaching Machines and the personalized learning they support on the structure of school teaching systems.The ana lysis of ChatGPT's teaching behaviors shows that in terms of teaching behaviors(activity and task),ChatGPT can carry out dialogue guidance,skill training,teaching demonstration,collaborative creation,interactive evaluation and other activities,but it lacks the superstructure of lessons,units,courses and so on.As the basis of its behavior,“action”also faces the challenge of low credibility and accuracy.Nevertheless,ChatGPT's cross-domain ability to understand and generate natural language still provides a new“ability base”for Intelligent Teaching System,which is expected to change the development mode of Intelligent Teaching System in a wide range,to improve its teach-ing performance,and to provide more comprehensive and effective support for personalized learning.Will there be structural changes in the school system as a result?This paper divides the question of“will or will not”into two questions:“can or can not”and“should or should not”,respectively in the level of fact logic and practical value.The analysis shows that the system structure of the existing school,such as subject courses,class teaching and teacher-led,remains stable and has not been greatly impacted by the existing Intelligent Teaching System and MOOCs.The teaching automation based on Large Language Models still needs to supplement the“superstructure”and consolidate the“basic actions”.For basic education,the elements of the school teaching system(such as teach-ers)may change considerably,but the system structure is likely to remain stable.
作者
张志祯
张玲玲
米天伊
丘诗萍
Zhizhen Zhang;Lingling Zhang;Tianyi Mi;Shiping Qiu
出处
《中国远程教育》
CSSCI
2023年第4期32-41,共10页
Chinese Journal of Distance Education
基金
2018年度教育部哲学社会科学研究重大课题攻关项目“世界主要国家教材建设研究”(项目编号:18JZD017)的研究成果。
作者简介
通讯作者:张志祯,北京师范大学教育学部教育技术基本理论研究中心副教授,zhangzz@bnu.edu.cn;张玲玲,北京开放大学国开业务部研究实习员;米天伊,北京师范大学教育学部教育技术学院硕士研究生;丘诗萍,北京师范大学教育学部教育技术学院硕士研究生。
