摘要
随着人工智能技术的飞速发展,以DeepSeek-R1为代表的推理模型为高等教育变革注入了新动力。在此背景下,文章首先梳理了从单一智能体到多智能体协同的技术演进历程,并分析了智能体赋能高等教育变革的可能场景与潜在优势,发现多智能体协同通过技术整合与策略优化,能够有效弥补单一智能体在面对复杂多变的教育场景时的局限性,实现教育场景中的复杂任务处理。之后,文章以“浙大先生”智能体平台为例,开展了多智能体协同的高等教育实践,并阐释了其在教育、科研、治理等场景进行的实践创新。最后,文章围绕“如何利用多智能体系统推动高等教育改革”的问题提出建议,以期为智能体赋能高等教育变革提供参考,并推动新质人才培养,助力实现高等教育的可持续、高质量发展。
With the rapid advancement of artificial intelligence technology,reasoning models represented by DeepSeek-R1 have injected new momentum into higher education transformation.Against this background,this paper first reviewed the technological evolution process from a single-agent to multi-agent collaboration,and analyzed the possible scenarios and potential advantages of agent-empowered transformation of higher education.It was found that multi-agent collaboration can effectively compensate for the limitations of a single-agent when facing complex and changeable educational scenarios through technological integration and strategic optimization,thereby enabling sophisticated task processing in educational scenarios.Subsequently,taking the agent platform named“Zhejiang University Great Master”as a case study,this paper carried out the higher education practice of multi-agent collaboration,and explained its practical innovations in scenarios such as education,scientific research,and governance.Finally,suggestions were proposed around the issue of“how to utilize multi-agent systems to drive higher education reform”,aiming to provide reference for agents to empower the transformation of higher education,promoting the cultivation of new quality talents,and contributing to the sustainable and high-quality development of higher education.
作者
陈默
杨玉辉
杨清元
陈娟娟
陈文智
CHEN Mo;YANG Yu-Hui;YANG Qing-Yuan;CHEN Juan-Juan;CHEN Wen-Zhi(Information Technology Center,Zhejiang University,Hangzhou,Zhejiang,China 310058;College of Education,Zhejiang University,Hangzhou,Zhejiang,China 310058)
出处
《现代教育技术》
2025年第5期111-118,共8页
Modern Educational Technology
基金
2023年国家自然科学基金面上项目“K12学生科学建模中多层次科学思维的智能评测及干预技术研究”(项目编号:62377042)的阶段性研究成果。
