AI掀起的新一轮智能化浪潮席卷全球，大众所熟悉的场景，是使用Agent协助完成工作任务，乘坐智能驾驶车辆出行，亦或观看机器人表演，静水流深的另 一面，是AI参与和驱动学术、科研，动力路径日益明晰。 12月23日，人才辈出的百年学府上海交通大学有位"特别成员"正式亮相——由上海交大联合华为，共同打造的"致远一号"智算平台，这标志着交大向"AI for Science"科研范式迈出了坚实一步。 其定位全国高校最大的国产智算算力基础设施，自2024年12月启动，用时不到一年，完成了千卡昇腾集群本地部署和云上千卡规模化应用，峰值算力633 PFLOPS、存储容量13P，令千亿参数级的大模型校内训练，落地成为现实。 它为师生提供的，是开箱即用的AI应用服务，为科研训练、教学实训与课程实验，提供了坚实的算力底座。 顺应第五范式的进化 致远一号的出现，恰逢其时。 在全球范围内，AI4Science即"人工智能驱动的科学研究"，正迅速占领高地，成为继经验科学、理论科学、计算科学和数据科学之后，现代科学活动的"第 五范式"。 有两点洞察，值得注意： 其一，越来越多科研人士、学术人士，积极拥抱AI、高效使用AI。 以算促学， 上 ...

Core Viewpoint - The emergence of AI-driven research paradigms signifies a fundamental transformation in scientific inquiry, shifting from human-centric methodologies to a collaborative human-machine approach, where AI plays an active role in the research process [2][6][21]. Group 1: Transformation of Research Paradigms - AI is driving a new research paradigm characterized by "data-intensive, intelligent emergence, and human-machine collaboration," fundamentally altering traditional research logic [1][2]. - The transition from traditional research methods to AI-driven approaches is seen as the emergence of a "fifth paradigm" in scientific research, emphasizing AI's role as an active participant rather than a mere tool [2][3]. - AI's capabilities allow for the generation of hypotheses and experimental designs, significantly increasing the efficiency of scientific discovery and altering the roles of researchers from creators to supervisors [3][4]. Group 2: Impacts on Scientific Discovery - AI's involvement in research has led to a model where AI proposes potential patterns or structures, which humans then validate, enhancing the efficiency of scientific discovery [4][5]. - The AlphaFold model exemplifies AI's potential to generate scientific knowledge, predicting protein structures with high accuracy, but emphasizes the need for rigorous validation of AI-generated predictions [3][4]. - The reliance on AI for hypothesis generation and experimental design may redefine the nature of scientific discovery, shifting from human-driven exploration to AI-guided inquiry [5][6]. Group 3: Challenges and Risks - The integration of AI in research raises concerns about the diminishing role of human researchers, as AI may take over a significant portion of intellectual labor, potentially impacting scientific knowledge transmission [11][13]. - The complexity of AI models may lead to challenges in understanding and validating AI-generated knowledge, creating a divide between human comprehension and AI outputs [13][14]. - Ethical considerations and the need for a robust governance framework are critical to ensure that AI's role in research does not undermine human oversight and accountability [19][20][21]. Group 4: Future Trends in Research - The future of scientific research is expected to be characterized by platform-based and collaborative approaches, integrating diverse data, algorithms, and models to enhance resource efficiency and output [14][15]. - AI is anticipated to evolve from a supportive tool to a comprehensive research partner, capable of autonomously conducting experiments and generating innovative theories [15][16]. - The democratization of research through AI may lower barriers to entry, allowing a broader range of individuals to engage in scientific inquiry, thus reshaping the landscape of research institutions [16][17]. Group 5: Skills and Education for Future Scientists - The core competencies for future scientists will shift towards problem definition, effective human-AI collaboration, and the ability to integrate and structure knowledge generated by AI [17][18]. - Educational frameworks will need to adapt, emphasizing critical thinking, problem-driven learning, and interdisciplinary approaches to prepare scientists for an AI-enhanced research environment [18][19].