ISSN 1671-3710
CN 11-4766/R
主办:中国科学院心理研究所
出版:科学出版社

• •    

AI导师与人类导师何以不同?差异效应与整合框架

李丽源, 吕虹洁, 褚福磊   

  1. 首都经济贸易大学工商管理学院, 北京 100070 中国
    北京师范大学经济与工商管理学院, 北京 100875 中国
  • 收稿日期:2026-06-09 修回日期:2026-10-01 接受日期:2026-11-13
  • 基金资助:
    教育部人文社科青年基金(25YJC630059)

How does AI mentoring differ from human mentoring? Differential effects and an integrative framework

Li Liyuan, Lv Hongjie, Chu Fulei   

  1. College of Business Administration, Capital University of Economics and Business 100070, China
    Business School, Beijing Normal University 100875, China
  • Received:2026-06-09 Revised:2026-10-01 Accepted:2026-11-13
  • Supported by:
    MOE (Ministry of Education in China) Project of Humanities and Social Sciences(25YJC630059)

摘要: 工智能(AI)导师作为一种新兴的发展支持来源,正在教育、医疗与组织培养等场景中承担部分传统导师功能。然而,现有研究多聚焦其技术应用与功能优势,对AI导师与人类导师效果差异的研究仍然有限。本文在明确AI导师概念边界的基础上,从任务与学习结果、关系与心理结果以及身份与规范结果三个维度,系统梳理AI导师与人类导师的差异效应。此外,本文进一步从认知、关系、以及身份与规范视角解释产生这些差异背后的理论机制,同时厘清个体、任务、以及系统与组织情境三方面的边界条件。未来研究应进一步澄清比较对象并加强直接比较,区分短期与长期结果,深化关系与身份机制研究,并关注边界条件、人机协同与本土化情境。

关键词: 人工智能导师, 人类导师, 导师制, 差异效应, 指导来源

Abstract: As an emerging source of developmental support, artificial intelligence (AI) mentors are increasingly assuming functions traditionally performed by human mentors across educational, healthcare, and organizational development contexts. However, existing research has primarily focused on the technological applications and functional advantages of AI mentors, while evidence regarding their comparative effectiveness relative to human mentors remains fragmented and limited. Building on a clear delineation of the conceptual boundaries of AI mentoring, this review systematically synthesizes the differential effects of AI and human mentors across three domains: task and learning outcomes, relational and psychological outcomes, and identity- and norm-related outcomes. Furthermore, we develop an integrative framework to explain why these differences emerge by examining underlying cognitive, relational, and identity-and-normative mechanisms. We also identify key boundary conditions at the individual, task, and system- and organizational-context levels that shape the effectiveness of AI versus human mentoring. Based on this synthesis, future research should strengthen direct comparative investigations, clarify comparison targets, differentiate short-term and long-term outcomes, advance understanding of relational and identity-related mechanisms, and further explore contextual contingencies, human-AI collaboration, and culturally embedded mentoring contexts. This review contributes to the emerging AI mentoring literature by providing a systematic account of the similarities and differences between AI and human mentors and by offering a foundation for future theoretical development and empirical inquiry.

Key words: AI mentor, human mentor, mentoring, differential effects, guidance source