典型科研场景下生成式人工智能使用的差异性分析——学科背景与人工智能素养的影响  

Analysis of Variability in the Use of Generative Artificial Intelligence in Typical Research Scenarios:The Influence of Disciplinary Background and AI Literacy

作　　者：李艳[1] 朱雨萌 孙丹[2] 许洁 翟雪松 LI Yan;ZHU Yumeng;SUN Dan;XU Jie;ZHAI Xuesong

机构地区：[1]浙江大学教育学院,浙江杭州310058 [2]杭州师范大学经亨颐教育学院,浙江杭州311121

出　　处：《现代远程教育研究》2025年第2期92-101,112,共11页Modern Distance Education Research

基　　金：2024年度国家自然科学基金面上项目“基于垂直领域大语言模型的智能中文写作平台设计、研发与应用研究”(62477040);浙江大学研究生教育研究课题“基于生成式人工智能(AI)的研究生科研训练及学术写作行为研究”(YJSJY20240101)。

摘　　要：以生成式人工智能(GenAI)为代表的智能技术推动了科研范式的转型,有助于复杂科研问题的解决。分析GenAI在科研场景中的使用差异及其影响因素,有利于高校智能化科研建设。基于对浙江大学1226位研究生的问卷调查发现:在科研中使用GenAI占比最多的前四个场景为文献翻译、信息检索、文献综述和润色降重;学科背景在信息检索、头脑风暴和代码生成三个场景中显著影响研究生使用GenAI;而人工智能素养在文献翻译、头脑风暴、文献综述、选择研究问题、实验设计、数据处理、代码生成和润色降重8个场景中显著影响研究生使用GenAI。此外,针对其中部分研究生的访谈分析发现:研究生认为能评估GenAI生成内容质量高低是合理使用GenAI的前提;GenAI应用技巧决定了GenAI辅助科研的效率;除学科背景和人工智能素养水平外,导师对GenAI的态度与使用水平以及高校GenAI资源开发与集成情况同样影响其在科研中使用GenAI。为促进研究生更好地利用GenAI赋能高质量科研创新,建议高校开发适配各专业教科研所需的垂直领域大模型,面向不同学科背景研究生设计融入GenAI使用的课程,全面提升高校师生人工智能素养。The advancement of intelligent technologies,represented by generative artificial intelligence(GenAI),has facilitated a paradigm shift in scientific research and contributed to the resolution of complex research problems.Analyzing the variability in the use of GenAI in research contexts and its influencing factors is essential for fostering the development of intelligent research in higher education institutions.Based on a survey of 1,226 graduate students at Zhejiang University,the study reveals that the four most frequently utilized research scenarios for GenAI are literature translation,information retrieval,literature review,and language refinement and text simplification.Disciplinary background significantly influences graduate students’use of GenAI in the scenarios of information retrieval,brainstorming and code generation.AI literacy exerts a significant impact on its use in literature translation,brainstorming,literature review,research question selection,experimental design,dataprocessing,code generation,and language refinement and text simplification.Furthermore,interviews with a subsetof graduate students indicate that the ability to assess the quality of GenAI-generated content is a prerequisite forits appropriate use;proficiency in GenAI application techniques determines its efficiency in supporting research.Inaddition to disciplinary background and AI literacy,factors such as advisors’attitudes toward and proficiency inGenAI,as well as universities’efforts in developing and integrating GenAI resources,also shape its adoption inresearch.To enhance graduate students’ability to leverage GenAI for high-quality research and innovation,it isrecommended that universities develop domain-specific large models tailored to the needs of various disciplines,design courses that integrate GenAI use based on students’disciplinary backgrounds,and comprehensivelyenhance AI literacy among faculty and students.

关 键 词：生成式人工智能 人工智能素养 科研场景 学科背景 

分 类 号：G434[文化科学—教育学]