| 1,204 | 2 | 101 |
| 下载次数 | 被引频次 | 阅读次数 |
随着技术进步，知识或技能本位的职业教育目标与评价导向已无法满足工作需求，亟需向职业行动能力转型。职业行动能力的整体性内涵及其作用机制，决定了其评价须同时实现整合能力观下的个性化评价、整体意义场下的表现性评价和多模态数据驱动的行动导向评价。但现有的先进评价方法高度依赖人工评分，难以落地和推广，传统人工智能等技术又受到难以攻克的“框架问题”和“映射问题”的限制，无法支持科学有效的自动化评价。生成式人工智能(GAI)作为语言世界的“具身者”和符号的“意义建构者”,能够在形式上模仿人类评分者所拥有的具身应对技能和隐性共识，使职业行动能力的自动化评价成为可能。研究基于“证据中心设计”思想，在确定学生模型的基础上，明确了任务模型的设计要求和证据模型的技术实现路径，从而构建了GAI赋能的职业行动能力评价框架。鉴于将GAI应用于职业行动能力评价会面临诸多伦理风险，研究进一步提供了相应的治理策略。
Abstract：With the development of technology, the knowledge-based or skill-based teaching objectives and assessment orientation of vocational education can no longer meet the needs of the workplace, and there is an urgent need to transform to professional action competence. The holistic connotation of professional action competence and the mechanism by which it works determines that its assessment must simultaneously achieve personalised assessment under the view of integrated competence, performance assessment under the field of holistic significance, and action-oriented assessment driven by multimodal data. However, the existing advanced assessment methods are highly dependent on manual scoring, which is difficult to be implemented and promoted, and traditional artificial intelligence and other technologies are limited by the insurmountable ‘frame problem' and ‘mapping problem', which cannot support scientific and effective automated assessment. Generative Artificial Intelligence(GAI), as the ‘embodiment' of the language world and the ‘meaning constructor' of symbols, can formally mimic the embodied coping skills and implicit consensus possessed by human raters, making it possible to automate the assessment of professional action competence. Based on the idea of ‘Evidence Centred Design', based on the identification of the student model, the study clarifies the design requirements of the task model and the technical implementation path of the evidence model, so as to construct a GAI-empowered professional action competence assessment framework. Since the application of GAI in professional action competence assessment will face many ethical risks, the study further provides the corresponding governance strategies.
[1]赵志群.构建适应数字时代的高技能人才培养培训体系[J].国家治理，2025(11):52-57.
[2]World Economic Forum.The Future of Jobs Report 2025[EB/OL].[2025-01-07].https://www.weforum.org/publications/the-future-of-jobs-report-2025/digest.
[3]赵志群.基于职业教育学理论学脉的技术技能人才培养新理念——新《职业教育法》学习心得[J].中国职业技术教育，2022(19):5-11.
[4]中华人民共和国中央人民政府.中华人民共和国职业教育法[EB/OL].[2022-04-21].https://www.gov.cn/xinwen/2022-04/21/content_5686375.htm.
[5]林君焕，黄广健，李金国，等.职业行动能力及评价的内涵探析、实证研究与教学启示——基于德国IHK结业考试制度本土化实践[J].中国职业技术教育，2023(32):73-87.
[6]KMK.Handreichung für die Erarbeitung von Rahmenlehrpl?nen der Kultusministerkonferenz für den berufsbezogenen Unterricht in der Berufsschule und ihre Abstimmung mit Ausbildungsordnungen des Bundes für anerkannte Ausbildungsberufe[EB/OL].[2021-06-16].https://www.kmk.org/fileadmin/veroeffentlichungen_beschluesse/2021/2021_06_17-GEP-Handreichung.pdf.
[7]中华人民共和国中央人民政府.中共中央国务院印发《教育强国建设规划纲要(2024—2035年)》[EB/OL].[2025-01-19].https://www.gov.cn/zhengce/202501/content_6999913.htm.
[8]赵志群，[德]菲利克斯·劳耐尔.COMET职业能力测评方法手册[M].北京：高等教育出版社，2018:54,57,70-73.
[9]Lievens F,Sackett P R,Dahlke J A,et al.Constructed response formats and their effects on minority-majority differences and validity[J].Journal of Applied Psychology,2019(5):715-726.
[10]俞鼎，李正风.生成式人工智能社会实验的伦理问题及治理[J].科学学研究，2024(1):3-9.
[11]卢国庆，杨沁，贺相春.生成式人工智能赋能高等教育形成性评价的价值、挑战及路径[J].电化教育研究，2024(11):84-91.
[12]Mertens D.Schlüsselqualifikationen,Thesen zur Schulung für eine moderne Gesellschaft[J].Mitteilungen aus der Arbeitsmarkt- und Berufsforschung,1974(7):36-43.
[13]徐朔.论“行动导向教学”的内涵和原则[J].职教论坛，2007(20):4-7.
[14]Reetz L.Zur Bedeutung der schlüsselqualifikationen in der Berufsbildung[C]// Schlüesselqualifikation-Fachwissen in der Krise.Hamburg:Feldhaus Verlag,1990:16-23.
[15]Laur-Ernst U.Schlüesselqualifikation-Innovative Ans?tze in den neugeordneten Berufen und ihre Konsequenzen für Lernen[C]// Schlüesselqualifikation-Fachwissen in der Krise.Feldhaus Verlag:Hamburg,1990:36-55.
[16]肖化移，李仲阳.论中职电脑动画设计专业教师能力标准的开发——一种基于功能分析法的研究[J].教师教育研究，2010(5):53-57.
[17]徐思.校企共评的高职金融专业学生职业能力评价体系研究[J].职教论坛，2017(17):32-35.
[18]刘金波.平台化：专业性学术期刊共同体建构研究[J].河南大学学报(社会科学版),2025(2):138-145+156.
[19]Gibson J J.The Ecological Approach to Visual Perception:classic edition[M].New York:Psychology Press,1986:18-19.
[20]Dreyfus H L.Overcoming the Myth of the Mental:How Philosophers Can Profit from the Phenomenology of Everyday Expertise[C]//Proceedings and Addresses of the American Philosophical Association.American Philosophical Association,2005(2):47-65.
[21]Engestr?m Y.Learning by Expanding:An Activity-Theoretical Approach to Developmental Research(Second Edition)[M].New York:Cambridge University Press,2014:70.
[22]赵志群.职业技术教育学[M].北京：中国人民大学出版社，2024:96-100.
[23]Engestr?m Y.Expansive learning at work:Toward an activity theoretical reconceptualization[J].Journal of Education and Work,2001(1):133-156.
[24][德]马库斯·加布里尔.为什么世界不存在[M].王熙，张振华，译.北京：商务印书馆，2022:67.
[25]Klerk S,Eggen J T H M,Veldkamp P B.A blending of computer-based assessment and performance-based assessment:Multimedia-Based Performance Assessment (MBPA).The introduction of a new method of assessment in Dutch Vocational Education and Training (VET)[J].Cadmo Giornale Italiano di Pedagogia Sperimentale,2014(1):39-56.
[26]Winther E,Achtenhagen F.Measurement of vocational competencies:a contribution to an international large-scale assessment on vocational education and training[J].Empirical research in vocational education and training,2009(1):85-102.
[27]OECD.PISA vocational education and training(VET):Assessment and analytical framework[R].Paris:OECD Publishing,2024.
[28]McCarthy J,Hayes P J.Some Philosophical Problems from the Standpoint of Artificial Intelligence[M]// Machine Intelligence 4.Edinburgh:Edinburgh University Press,1969:463-502.
[29]尚杰.从语言的边界到显露的图像[J].中国社会科学，2022(11):49-68+205.
[30]徐静，骆方，马彦珍，等.开放式情境判断测验的自动化评分[J].心理学报，2024(6):831-844.
[31]Meissner R,P?gelt A,Ihsberner K,et al.LLM-generated competence-based e-assessment items for higher education mathematics:methodology and evaluation[J].Frontiers in Education,2024(9):1427502.
[32]Patel N,Nagpal P,Shah T,et al.Improving mathematics assessment readability:Do large language models help?[J].Journal of Computer Assisted Learning,2023(3):804-822.
[33]Li J,Jangamreddy N K,Hisamoto R,et al.AI-assisted marking:Functionality and limitations of ChatGPT in written assessment evaluation[J].Australasian Journal of Educational Technology,2024(4):56-72.
[34]Jauhiainen J S,Guerra A G.Generative AI in education:ChatGPT-4 in evaluating students’ written responses[J].Innovations in Education and Teaching International,2025(4):1377-1394.
[35]Lee U,Kim Y,Lee S,et al.Can we use GPT-4 as a mathematics evaluator in education?:Exploring the efficacy and limitation of LLM-based automatic assessment system for open-ended mathematics question[J].International Journal of Artificial Intelligence in Education,2025(3):1560-1596.
[36]Jiang B W,Xie Y X Y,Hao Z Q,et al.A Peek into Token Bias:Large Language Models Are Not Yet Genuine Reasoners[C]// 2024 Conference on Empirical Methods in Natural Language Processing.Stroudsburg:ACL,2024:4722-4756.
[37]de Winter J C F,Dodou D,Stienen A H A.ChatGPT in Education:Empowering Educators through Methods for Recognition and Assessment[J].Informatics-Basel,2023(4):1-14.
[38]Estévez-Ayres I,Callejo P,Hombrados-Herrera M á,et al.Evaluation of LLM Tools for Feedback Generation in a Course on Concurrent Programming[J].International Journal of Artificial Intelligence Education,2025(2):774-790.
[39]Sun Z Q,Shen Y K,Zhou Q H,et al.Principle-Driven Self-Alignment of Language Models from Scratch with Minimal Human Supervision[C]// Advances in Neural Information Processing Systems 36.California:NIPS,2023,36:2511-2565.
[40]余越凡，赵志群.AIGC时代数字人赋能人机协同学习：价值、框架与路径[J].远程教育杂志，2025(2):45-52.
[41]宇璐，潘海英.基于语言模因论的语言接触过程与结果探析[J].东北师大学报(哲学社会科学版),2015(5):181-186.
[42]余越凡，赵志群，黄方慧.基于大模型的职业教育会话机器人构建与应用[J].电化教育研究，2025(8):65-72.
[43][美]埃贡·G.古贝，伊冯娜·S.林肯.第四代评估[M].秦霖，蒋燕玲，等，译.北京：中国人民大学出版社，2008:106-109.
[44]Liu Y,Lin K Q,Chen C W,et al.VideoMind:A Chain-of-LoRA Agent for Long Video Reasoning[EB/OL].[2025-03-31].https://arxiv.org/abs/2503.13444.
[45]Zhu Y L,Moniz J R A,Bhargava S,et al.Can Large Language Models Understand Context?[C]// Findings of the Association for Computational Linguistics:EACL 2024.Stroudsburg:ACL,2024:2004-2018.
[46]Gupta O,Marrone S,Gargiulo F,et al.Understanding Social Biases in Large Language Models[J].AI,2025(5):1-26.
[47]Mislevy R J,Almond R G,Lukas J F.A brief introduction to evidence-centered design[R].New Jersey:Research & Development Division Princeton,2023.
[48]朱莎，吴砥，杨浩，等.基于ECD的学生信息素养评价研究框架[J].中国电化教育，2020(10):88-96.
[49]国家职业分类大典修订工作委员会.中华人民共和国职业分类大典(2022年版)[M].北京：中国劳动社会保障出版社，2022:36-37.
[50]Rauner F.Handbook of Fundamentals of Modern Vocational Education[M].Singapore:Springer,2024:732-733.
[51]赵志群，高帆.职业院校学生职业能力发展现状及其影响因素——五类制造业专业基于COMET测评数据的分析[J].中国电化教育，2022(6):47-55.
[52]赵志群，王炜波.德国职业教育设计导向的教育思想研究[J].中国职业技术教育，2006(32):62-64.
[53]赵志群.关于行动导向的教学[J].职教论坛，2008(20):1.
[54]姜大源.论高等职业教育课程的系统化设计——关于工作过程系统化课程开发的解读[J].中国高教研究，2009(4):66-70.
[55]徐涵.学习领域课程与项目课程的比较研究[J].职教论坛，2015(15):13-17.
[56]朱伟强，崔允漷.基于标准的课程设计：开发表现性评价[J].全球教育展望，2007(10):43-48.
[57]赵志群，孙钰林，罗喜娜.“1+X”证书制度建设对技术技能人才评价的挑战——世界技能大赛试题的启发[J].中国电化教育，2020(2):8-14.
[58]郑勤华，陈丽，柴唤友，等.基于信息技术的表现性评价：内涵、作用点与发展路向[J].中国电化教育，2023(3):55-61.
[59]余明华，张治，刘小龙，等.基于xAPI的研究性学习多源数据融合研究[J].现代远距离教育，2022(3):63-69.
[60]Rosso P,Yang D Q,Cudré-Mauroux P.Beyond Triplets:Hyper-Relational Knowledge Graph Embedding for Link Prediction[C]// Web Conference 2020:Proceedings of the World Wide Web Conference.New York:ASSOC Computing Machinery,2020:1885-1896.
[61]Panda P,Agarwal A,Devaguptapu C,et al.HOLMES:Hyper-Relational Knowledge Graphs for Multi-hop Question Answering using LLMs[C]// Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics.Stroudsburg:ACL,2024(1):13263-13282.
[62]王佑镁，房斯萌，柳晨晨.风险社会视角下教育人工智能伦理风险分类框架研究[J].现代远距离教育，2024(3):28-37.
[63]陈新汉.关于评价论研究的几个问题之我见[J].天津社会科学，2000(2):31-35.
①例如，婴儿反复将手伸向杯子时，不是应用“抓握”概念，而是通过知觉材料的动态整合形成“抓握规范”。后续，杯子的知觉特征将直接向个体发出行动邀请，无需概念翻译。Dreyfus认为，成人、婴儿和动物拥有相同的应对技能，成人即使需要经历概念指导行动的过程，但在技能成熟后，概念将隐退，这种应对的满足条件仍是规范性的而非概念性的。
②此处的“动机”不是心理学概念，而是对可供性引发行动的方式的命名，加上身体获取最大把握(Maximal Grip)的倾向构成了“动机空间”,由Merleau-Ponty提出，Dreyfus继承了这一概念。
③语言学中，“能指”指符号可感知的物理形态，“所指”指符号所代表的心理概念或实际对象。
④“模因”指人类社会文化传递的复制因子，其表现形态涵盖词语、图像、格调、表情等所有形式。“模因复合体”则指由多个相互关联的模因组成的认知集合体，这些模因通过协同作用形成稳定的传播结构。
基本信息:
DOI：10.13927/j.cnki.yuan.20250908.001
中图分类号:TP18;F49
引用信息:
[1]余越凡,赵志群.生成式人工智能赋能职业行动能力评价：原理与框架[J].现代远距离教育,2025,No.221(05):21-32.DOI:10.13927/j.cnki.yuan.20250908.001.
基金信息:
2024年度国家重点研发计划“智能交互实验教学关键技术研究及应用”子课题“多模态多传感器融合的实验过程跟踪分析与智能评测”(编号：2024YFC3308504); 2024年度教育部人文社会科学研究青年基金项目“基于多模态学习分析的职教高考技能评价实证研究”(编号：24YJC880054)