【作 者】：
汪维富，毛美娟， 余 慧，孙如意
【关 键 词 】：
人工智能； 大语言模型； 生成式人工智能； 第五范式； 计算再现性
【栏 目】：
理论探讨
【中文摘要】：
新一代人工智能正在引领科学研究方法与知识发现逻辑的颠覆式创新，推动教育研究向人工智能驱动的第五研究范式迈进，然而，此新范式尚处于理论探索的初级阶段。通过深入剖析，研究阐明了新范式的发轫逻辑：借助深度学习算法与贝叶斯思维，人工智能在科学实践中扮演着研究设计的预判先知者、合成样本数据的智能代理者、知识自主涌现的推动者等独特角色。人工智能通过促进跨学科协作研究、推动科研组织模式变革、拓展知识生产主体范围、回归实践价值取向、促成新旧范式有机统合等，深刻影响着知识发现逻辑与科学研究核心议程。然而，新范式发展也面临着机器生成幻觉知识、挤压人类研究主体发挥空间、计算再现性困难、研究者人工智能素养普遍不高等挑战。为此，教育研究共同体需搭建适用于特定领域的开源大模型、维护人类研究的主体性与首要地位、构建适应新范式的科研伦理规范、强化知识发现核验制度建设、培育研究者算法意识与人工智能素养。
【英文摘要】：
The new generation of artificial intelligence is leading disruptive innovations in scientific research methods and the logic of knowledge discovery, promoting educational research to move towards the fifth paradigm driven by artificial intelligence. However, this new paradigm is still in its nascent stage of theoretical exploration. Through an in-depth analysis, this study elucidates the incipient logic of the new paradigm: with the aid of deep learning algorithms and Bayesian thinking, artificial intelligence plays unique roles in scientific practice, such as the predictive precursors of research design, the intelligent agents of synthetic sample data, and the promoter of autonomous knowledge emergence. By fostering interdisciplinary collaborative research, driving the transformation of scientific research organization models, expanding the scope of knowledge production entities, reorienting towards practical value, and facilitating the organic integration of old and new paradigms, artificial intelligence profoundly influences the logic of knowledge discovery and the core agendas of scientific research. Nevertheless, the development of this new paradigm also confronts challenges such as machine-generated illusionary knowledge, the marginalization of human researchers' roles, difficulties in computational reproducibility, and the generally low level of researchers' artificial intelligence literacy. To address these issues, the educational research community needs to take the following measures: to develop open-source large models tailored to specific fields, maintain the subjectivity and primary position of human researchers, establish scientific research ethical norms adapted to the new paradigm, strengthen the construction of knowledge discovery and verification systems, and to cultivate researchers' algorithmic awareness and artificial intelligence literacy.