«上一篇/Previous Article|本期目录/Table of Contents|下一篇/Next Article»

清华大学教育研究[ISSN:1001-4519/CN:11-1610/G4]

卷:

期数:

2026年05期

页码:

19-28

栏目:

人工智能与教育

出版日期:

2026-10-20

文章信息/Info

Title:

Competencies for Talent Development in the Age of Arti?cial Intelligence ——A Large-Scale Literature Analysis Using Semantic Embeddings

作者:

张羽¹ 李金赫² 于济凡¹

1. 清华大学 教育学院；2. 马来西亚理科大学 教育学院

Author(s):

ZHANG Yu¹ LI Jin-he² YU Ji-fan¹

1. School of Education, Tsinghua University;2. School of Educational Studies, Universiti Sains Malaysia

关键词:

高等教育; 人工智能; 人才素养; 人才培养; 语义嵌入

Keywords:

higher education;  arti?cial intelligence;  competencies;  talent development;  semantic embedding

分类号:

G642

DOI:

10.14138/j.1001-4519.2026.05.001910

文献标志码:

A

摘要:

人工智能的快速发展要求高等教育重新审视人才素养的构成及其培养方式。本文采用语义嵌入辅助的多源文献分析，通过语义分组组织材料比较，结合原文核查与概念综合，将人工智能时代的人才素养要求归纳为“智能技术理解与运用、知识探究与专业判断、自主学习与自我发展、审美体验与文化表达、社会参与与公共责任”五个核心维度，涵盖十三项具体内容；同时，讨论以“尝试与改进”为素养的实践取向，说明不同素养要求如何在实际活动中展开并相互联系，将培养内容与学习经历相联系。本文基于跨领域海量文献，采用透明的语义计算和可复现的分析过程，研究结论具有更强的普适性，为高等教育培养目标的重新确认、课程体系改革与育人实践改进提供参考。

Abstract:

The rapid development of artificial intelligence calls for higher education to reconsider the competencies required for talent development and how such competencies should be cultivated. This study conducts a semantic embedding-assisted analysis of multi-source literature. By organizing cross-textual comparisons through semantic grouping and combining original source veri?cation with conceptual synthesis,it identi?es ?ve core dimensions of competencies for talent development in the age of arti?cial intelligence:understanding and use of intelligent technologies; knowledge inquiry and professional judgment; self-directed learning and self-development; aesthetic experience and cultural expression; and social participation and public responsibility. These dimensions encompass thirteen speci?c components. The study further discusses a practice-oriented perspective centered on “trying and improving,” explaining how different competency requirements are enacted and interconnected in practical activities and linking educational content with learning experiences. Drawing on massive cross-disciplinary literature and employing transparent computational textprocessing and a reproducible analytical process, the study yields highly generalizable ?ndings. These ?ndings provide a reference for reaffirming educational goals, reforming curricula, and improving the educational practices in higher education.

常用功能

更新日期/Last Update: 2026-10-20